Method for calculating annual power generation time of overtopping wave energy converter based on ERA5 data and wave height conversion

Abstract

The present application belongs to the cross field of wave energy power generation technology field and numerical simulation of ocean engineering, and discloses an OWC wave energy power station annual power generation time length calculation method based on ERA5 data and specific wave height conversion, comprising: extracting offshore reference point hourly wave element data from ERA5 global reanalysis data set; inputting the data into a spatial physical information neural network conversion model along the wave propagation path, and outputting the effective wave height time sequence of the control point in front of the dike; determining the offshore wave power generation state according to the preset power generation threshold of the unit; determining the wind wave power generation state based on the wind speed, wind direction and continuous duration constituted wind wave proxy criterion, and merging the two hourly logical ors; and according to the merged power generation state sequence, the annual comprehensive equivalent power generation days are counted. Thus, the neural network with physical constraint enhancement realizes high-precision wave height conversion, and through the hourly union rule, repeated statistics is avoided, which is suitable for economic evaluation and site selection demonstration of port caisson type wave energy power generation device.

Description

Technical Field

[0001] This invention belongs to the interdisciplinary field of wave energy power generation technology and marine engineering numerical simulation. Specifically, it relates to a method for estimating the annual power generation duration of OWC wave energy power stations based on ERA5 data and wave height conversion. More specifically, it relates to a method for estimating the annual power generation duration of OWC (Oscillating Water Column) wave energy power stations based on multi-source data fusion and physical information neural network (PINN). It is applicable to the economic evaluation and site selection demonstration of port caisson wave energy power generation devices. Background Technology

[0002] As a clean and renewable energy source, wave energy resource assessment and power generation potential prediction are core bases for wave power plant project decisions. OWC (Oscillating Water Column) wave energy devices, due to their simple structure and high reliability, have become one of the mainstream technologies for wave energy power generation. However, in engineering practice, the calculation of the annual power generation duration of OWC power plants faces the following technical challenges: (1) There is a complex conversion relationship between the wave conditions in the open sea and in front of the breakwater. Waves propagate from the open sea to the breakwater in the harbor, and are affected by factors such as topography, diffraction, refraction, and shallow water deformation, resulting in significant changes in wave height, wave direction, and wave period. Wave observation data in the open sea cannot be directly used to assess the actual power generation potential of the equipment in front of the breakwater. A conversion relationship needs to be established between the two through numerical wave models or measured calibration. However, the conversion coefficients themselves vary nonlinearly with multiple factors such as wave direction, wave height, and period. The accuracy of traditional empirical formulas or linear interpolation methods is limited, resulting in a large nonlinear error in the conversion coefficients.

[0003] (2) Long-term field measurement data acquisition is costly. Accurate calculation of annual power generation duration requires continuous observation data covering different seasons and wave conditions throughout the year. Field measurements are not only costly and time-consuming, but also difficult to guarantee data continuity due to adverse sea conditions.

[0004] (3) The waves from the open sea and the wind-generated waves in the harbor are prone to double counting. When calculating the effective power generation time, the existing methods often calculate the contribution of the waves from the open sea and the contribution of the wind-generated waves in the harbor separately and then simply add them together. However, the two types of events often overlap in time, and directly adding them together will cause double counting and overestimate the actual power generation time. Summary of the Invention

[0005] This invention addresses the problems of insufficient accuracy of conversion factors and overlapping counting of incoming waves and wind-generated waves in existing OWC wave energy power station annual power generation duration estimation methods. It proposes a new method based on ERA5 data and wave height conversion. This method constructs a spatial PINN model along the wave propagation path, embedding energy conservation physical constraints into the training process. It also employs hourly logical union rules to handle overlapping events of incoming waves and wind-generated waves. Specifically, it organically integrates offshore wave data, numerical wave simulation, limited field measurements, and hourly union statistics of wind and waves to form a systematic and reusable annual power generation duration estimation method, achieving statistical estimation of the equivalent effective power generation duration over a long time series.

[0006] In a first aspect, the present invention provides a method for estimating the annual power generation duration of an OWC wave energy power station based on ERA5 data and wave height conversion, comprising the following steps: Hourly wave element data of offshore reference points in the engineering area are extracted from the ERA5 global reanalysis dataset. The wave element data includes significant wave height, mean period, and main wave direction. The hourly wave element data is input into a pre-constructed and trained spatial physical information neural network conversion model along the wave propagation path, and the effective wave height time series of the control points in front of the dike is output. Based on the preset power generation threshold wave height of at least two wave energy generator units, it is determined hourly whether the incoming waves from the ocean cause each unit to be in an effective power generation state, and the effective power generation state of any one unit is taken as the effective power generation state of the power station, thus obtaining the hourly effective power generation state sequence corresponding to the incoming waves from the ocean. Based on the wind-forming wave power generation proxy criteria consisting of wind speed, wind direction and continuous duration, the wind-forming wave power generation status in the port is determined hourly. The time period corresponding to the wind-forming wave power generation status in the port is then logically ORed with the hourly effective power generation status sequence corresponding to the incoming waves from the open sea to obtain the comprehensive effective power generation status sequence of the power station. The annual equivalent number of power generation days is calculated based on the comprehensive effective power generation status sequence of the power station.

[0007] Furthermore, the construction method of the spatial physical information neural network conversion model includes: dividing at least two wave direction intervals according to the incident wave direction in the outer sea; for each wave direction interval, a corresponding propagation path is preset between the outer sea reference point and the control point in front of the dike; the propagation path is discretized into multiple path nodes along its arc length coordinate; the spatial physical information neural network conversion model takes the propagation path arc length coordinate, the significant wave height of the outer sea reference point, the average period of the outer sea reference point, and the sine and cosine values ​​of the main wave direction of the outer sea reference point as inputs, and the significant wave height of each node along the current propagation path as output; wherein, the node at the end of the propagation path corresponds to the control point in front of the dike, and the significant wave height at this node is the significant wave height of the control point in front of the dike.

[0008] Furthermore, the spatial physical information neural network conversion model is a multi-layer fully connected neural network. Its input layer receives the arc length coordinates of the propagation path, the effective wave height of the offshore reference point, the average period of the offshore reference point, and the sine and cosine values ​​of the main wave direction of the offshore reference point. After processing by multiple fully connected hidden layers, the output layer outputs the effective wave height of each node along the current propagation path.

[0009] Furthermore, the training process of the space physics information neural network conversion model incorporates the physical constraint of conservation of wave energy flow along the propagation path. The physical constraint is constructed based on the physical law that balances the change of wave energy flow along the propagation path with the dissipation term. The dissipation term includes bottom friction dissipation rate and wave breaking dissipation rate. The physical loss is the weighted summation of the residuals of wave energy flow along the propagation path at each physical configuration point.

[0010] Furthermore, the overall loss of the space physics information neural network conversion model is composed of a weighted average of data loss, physical loss, and boundary loss. The data loss is used to constrain the consistency between the model output and the measured data and numerical simulation results. The physical loss is used to constrain the model output to satisfy the conservation of wave energy flow along the path. The boundary loss is used to constrain the model output at the offshore reference point to be consistent with the input wave height. Moreover, a gradient alignment adaptive weight adjustment strategy is adopted to dynamically adjust the weights of data loss and physical loss. When a conflict in the gradient directions of the two is detected, the weight of physical loss is reduced and the weight of data loss is increased.

[0011] Furthermore, the method for determining whether the incoming waves from the open sea cause each unit to be in an effective power generation state based on the preset power generation threshold wave height of at least two wave energy generator units is as follows: the power generation threshold wave height of the first unit and the power generation threshold wave height of the second unit are used to determine the hourly power generation state of the two units respectively, and the comprehensive effective power generation state of the power station from the open sea is the logical OR of the power generation state of each unit; the effective power generation hours of each wave direction interval and each month are counted, and the total effective power generation hours are divided by 24 to convert into the equivalent effective power generation days.

[0012] Furthermore, in the wind-induced wave power generation proxy criterion, the wind speed threshold, wind direction range, and duration threshold are calibrated based on the effective wind distance of the port area, on-site wind condition records, and device operation records; when determining the wind-induced wave power generation status within the port hourly, when the hourly wind speed reaches the wind speed threshold, the hourly wind direction falls into the wind direction range, and the continuous duration reaches the duration threshold, the corresponding time is determined to be the wind-induced wave power generation status within the port.

[0013] Furthermore, the comprehensive effective power generation state sequence of the power station is obtained by performing an hourly logical OR operation on the hourly effective power generation state sequence corresponding to the wave-generated power generation state in the open sea and the time period corresponding to the wave-generated power generation state in the harbor; the comprehensive equivalent power generation days for the whole year are obtained by summing the comprehensive effective power generation state sequence of the power station hourly and dividing by 24, and the overlapping time period is directly determined by the hourly state sequence.

[0014] Furthermore, it also includes: calculating hourly power based on the effective wave height time series of the control point in front of the dike and the wave-to-electric power conversion characteristics of the wave energy power generation device; multiplying the hourly power by the time step and summing them up to obtain the monthly power generation. The hourly power is calculated in the following way: determining the hourly output of each unit based on the wave height-period-power interpolation function of each unit, using the effective wave height and average period in front of the dike, and limiting the output based on the rated power of each unit; the instantaneous total output of the power station is the sum of the outputs of each unit, and the overall output is limited by the upper limit of the grid-connected feeder capacity of the power station; multiplying the hourly power after overall limitation by the time step and summing them up to obtain the monthly power generation.

[0015] Furthermore, it also includes: calculating hourly power based on the effective wave height time series of the control point in front of the dike and the wave-to-electric conversion power characteristics of the wave energy power generation device; multiplying the hourly power by the time step and summing them up to obtain the monthly power generation. The hourly power is calculated as follows: calculating the wave energy capture power of each unit based on the wave-to-electric comprehensive conversion efficiency function and equivalent capture width of each unit, combined with the effective wave height and wave group velocity in front of the dike, and limiting the amplitude based on the rated power of each unit; the instantaneous total output of the power station is the sum of the output of each unit, and the overall amplitude is limited by the upper limit of the grid-connected feeder capacity of the power station; multiplying the hourly power after overall amplitude limitation by the time step and summing them up to obtain the monthly power generation.

[0016] In a second aspect, a control device is provided, including a processor and a storage device, the storage device being adapted to store a plurality of program codes, the program codes being adapted to be loaded and executed by the processor to perform the method of the first aspect.

[0017] In a third aspect, a computer-readable storage medium is provided having a plurality of program codes stored thereon, the program codes being adapted to be loaded and run by a processor to perform the method described in the first aspect.

[0018] Compared with the prior art, the present invention has the following advantages: (1) Physical constraints have a clear calculation process. Spatial PINN along the propagation path makes the partial derivatives have a clear calculation object. The water depth is constructed as a continuous differentiable function through cubic spline interpolation. The physical constraint boundary is clear, avoiding the problem of losing chain rule terms in automatic differentiation caused by traditional PINN input design.

[0019] (2) The time-by-time logical union rule is used to replace the simple arithmetic addition, so that when the offshore wave power generation state and the wind wave proxy criterion are met at the same time, it is only counted once, thus avoiding repeated accumulation from the statistical rules.

[0020] (3) The roles of multiple data sources are clear. SWAN simulation results serve as auxiliary training samples, and field measured data serve as calibration samples. Physical constraints are independent of both, and the three do not overlap and are logically consistent.

[0021] (4) Set the propagation path according to the wave direction interval. Different incident wave directions in the open sea correspond to different refraction paths and water depth profiles. The zoning setting makes the calculation model more in line with the actual wave propagation process. Attached Figure Description

[0022] The disclosure of this invention will become more readily understood with reference to the accompanying drawings. It will be readily understood by those skilled in the art that these drawings are for illustrative purposes only and are not intended to limit the scope of protection of this invention. Furthermore, similar numbers in the drawings are used to denote similar components, wherein: Figure 1 This is a flowchart illustrating the method for calculating the annual power generation duration of an OWC wave energy power station based on ERA5 data and wave height conversion according to an embodiment of the present invention. Figure 2 This is a schematic diagram of the structure of the spatial physical information neural network conversion model along the wave propagation path according to an embodiment of the present invention; Figure 3 This is a logical diagram illustrating the hourly union correction of incoming waves from the open sea and wind-generated waves within the harbor, according to an embodiment of the present invention. Detailed Implementation

[0023] Some embodiments of the present invention will now be described with reference to the accompanying drawings. Those skilled in the art should understand that these embodiments are merely illustrative of the technical principles of the present invention and are not intended to limit the scope of protection of the present invention.

[0024] In the description of this invention, "module" and "processor" can include hardware, software, or a combination of both. A module can include hardware circuitry, various suitable sensors, communication ports, memory, and may also include software components, such as program code, or a combination of software and hardware. A processor can be a central processing unit, microprocessor, image processor, digital signal processor, or any other suitable processor. The processor has data and / or signal processing capabilities. The processor can be implemented in software, in hardware, or a combination of both. Non-transitory computer-readable storage media includes any suitable medium capable of storing program code, such as magnetic disks, hard disks, optical disks, flash memory, read-only memory, random access memory, etc. The term "A and / or B" means all possible combinations of A and B, such as only A, only B, or A and B. The terms "at least one A or B" or "at least one of A and B" have a similar meaning to "A and / or B" and can include only A, only B, or A and B. The singular terms "a" or "this" can also include plural forms.

[0025] In this invention, wave height conversion refers to the process of converting the effective wave height of an offshore reference point into the effective wave height of a control point in front of the seawall, based on the energy dissipation and deformation characteristics during wave propagation. Essentially, it establishes a nonlinear mapping relationship between offshore wave height and seawall wave height. The wave height conversion model is a computational model that realizes this mapping relationship. In this field, wave height conversion can also be understood as the calculation process of wave height transfer ratio or wave height conversion coefficient.

[0026] Example 1: Figure 1 This is a flowchart illustrating the method for calculating the annual power generation duration of an OWC wave energy power station based on ERA5 data and wave height conversion according to an embodiment of the present invention. Figure 1 As shown, the method for estimating the annual power generation duration of an OWC wave energy power station based on ERA5 data and wave height conversion includes the following steps: S1: Extract hourly wave element data of the offshore reference point of the engineering sea area from the ERA5 global reanalysis dataset. The wave element data includes significant wave height, mean period and main wave direction. Step S1 is used to extract wave data based on the ERA5 global reanalysis dataset.

[0027] Using the ERA5 global reanalysis dataset from the European Centre for Medium-Range Weather Forecasts (ECMWF), hourly wave data, including significant wave height, were extracted from reference points offshore the engineering area over a period of more than 10 years. Average period Main wave direction The extracted data were analyzed by direction and month to establish a long-term statistical distribution of offshore waves.

[0028] S2: Input the hourly wave element data into the pre-constructed and trained spatial physical information neural network conversion model along the wave propagation path, and output the effective wave height time series of the control points in front of the dike; Step S2 is used to construct a PINN (Physics-Informed Neural Networks) conversion model along the wave propagation path.

[0029] Traditional methods rely solely on numerical wave models (such as SWAN, Simulating Waves Nearshore) to calculate the wave height conversion factor from the open sea to the breakwater, making it difficult to effectively utilize limited field measurement data for calibration.

[0030] This invention addresses the wave propagation problem and includes specific improvements to PINN, comprising the following components: a one-dimensional spatial PINN architecture along the propagation path, independent paths for wave direction intervals, cubic spline depth interpolation, and gradient alignment adaptive weight adjustment. Specifically, the spatial PINN model (hereinafter referred to as the "spatial PINN model") is constructed as follows: At least two wave direction intervals are divided based on the incident wave direction in the open sea. For each wave direction interval, a corresponding propagation path is preset between the offshore reference point and the control point in front of the dike. The propagation path is discretized into multiple path nodes along its arc length coordinates. The spatial PINN model takes the propagation path arc length coordinates, the significant wave height of the offshore reference point, the average period of the offshore reference point, and the sine and cosine values ​​of the main wave direction of the offshore reference point as inputs, and the significant wave heights of each node along the current propagation path as outputs.

[0031] like Figure 2 The diagram shown is a schematic representation of the spatial physical information neural network conversion model along the wave propagation path according to an embodiment of the present invention. The architecture of the spatial PINN model is as follows: (a) Wave direction interval and propagation path settings: Based on the direction of the incident wave in the open sea, the sea wave condition is divided into at least two wave direction intervals. For each wave direction interval... Pre-set corresponding propagation paths between the offshore reference point and the control point in front of the dike. The propagation path Along its arc length coordinates Discretized into multiple path nodes, where Corresponding offshore reference points, Corresponding control points in front of the dike, For the first Each wave direction interval corresponds to the length of the propagation path.

[0032] For any given moment, the outer sea wave direction Determine its wave direction interval And the wave propagation path corresponding to the interval , water depth function along the path And the PINN model with dissipative parameters in the input space.

[0033] (b) Network Inputs and Outputs: The spatial PINN model is a multi-layer fully connected neural network. Its input layer receives the arc length coordinates of the propagation path, the significant wave height of the reference point at the outer sea, the average period of the reference point at the outer sea, and the sine and cosine values ​​of the main wave direction of the reference point at the outer sea. After processing by multiple fully connected hidden layers, the output layer outputs the significant wave height of each node along the current propagation path. The effective wave height at the end of the propagation path is the effective wave height of the control point in front of the dike. Specifically: Neural network backbone input for:

[0034] in, The arc length coordinates of the propagation path (satisfying) ), These are the effective wave height, mean period, and main wave direction of the offshore reference point, respectively.

[0035] The network output is the effective wave height of each node along the current propagation path. The end point of the propagation path ( The node at () is the control point in front of the dike, and the corresponding output in front of the dike is ,in, The effective wave height of the control point in front of the dike is predicted by the model, and the subscript i indicates the front of the dike.

[0036] water depth The physical residual calculation module is introduced as an explicit function of the known path geometry and terrain, and does not directly participate in the neural network input. For the discrete water depth node data extracted from each propagation path, cubic spline interpolation is used to construct a continuous and first-order differentiable water depth function. This enables the automatic differentiation mechanism to process the output of the neural network. The water depth function obtained by differentiable interpolation They jointly participate in the calculation of group velocity and wave energy flow, and calculate the spatial derivative through automatic differentiation.

[0037] (c) Physical constraint embedding: The training process of the space physics information neural network conversion model incorporates the physical constraint of conservation of wave energy flow along the propagation path. This physical constraint is constructed based on the physical law that balances the change in wave energy flow along the propagation path with the dissipation term. The dissipation term includes bottom friction dissipation rate and wave breaking dissipation rate. The physical loss is a weighted sum of the residuals of wave energy flow along the propagation path at each physical configuration point. Specifically: Construct physical residuals based on the conservation of energy flow along the wave propagation process. For the first... A propagation path, along which wave energy flows Defined as:

[0038] in, Seawater density, unit: ; Gravitational acceleration, unit: . The wave group speed.

[0039] Wave group speed Calculated based on the linear wave dispersion relation:

[0040]

[0041] in, Wave angular frequency ( , (meaning the average period) For wave number, according to the dispersion relation Seeking, This represents the local water depth (i.e., the water depth at that location). Wave energy flow residual. Defined as the sum of the change in energy flow along the path of the wave and the dissipation term:

[0042] in, To achieve the lowest frictional dissipation rate, The wave breaking dissipation rate is represented by the bottom friction dissipation parameters and the breaking dissipation parameters, which are adopted from the parameterized model publicly available in SWAN. These parameters can be determined based on the seabed type, wave breaking characteristics, or existing engineering configurations in the target sea area, and their specific values ​​are not limited. In steady-state energy conservation, the energy flow along the wave path is... (Energy decays along the path), and a positive dissipation term. The sum tends to 0.

[0043] Physical loss The weighted summation of the residuals at all physical placement points along the propagation path is used to constrain the model output to satisfy the conservation of energy flow along the path of the wave.

[0044] in, This represents the number of propagation paths corresponding to the wave direction interval. For the first The number of physical configuration points along the propagation path For the first The first of the transmission paths Arc length coordinate sampling points .

[0045] The energy flow conservation constraint along the propagation path is used to constrain the wave height variation trend on the preset equivalent propagation path; the port plane diffraction, local reflection and complex shoreline shielding effects are calibrated by SWAN simulation auxiliary samples and on-site measured wave height data in front of the dike, and the one-dimensional physical constraint is not used as a model to independently solve the entire two-dimensional wave field in the port.

[0046] (d) Boundary constraints and data constraints: Reference point at sea Apply boundary constraints at the boundary, and the boundary loss This is used to ensure that the model output at the offshore reference point is consistent with the input wave height;

[0047] in, This represents the total number of boundary sampling points; Indexing of boundary sampling points ; The model outputs the significant wave height at the m-th boundary sampling point, at the reference point s=0 in the open sea, along the q-th propagation path; Input the significant wave height for the offshore reference point at the m-th boundary sampling point.

[0048] The data constraints are composed of both field-measured wave height data in front of the dike and SWAN numerical simulation results, with data loss... This is used to constrain the consistency between the model output and the measured data and numerical simulation results. To clarify the calibration effect of data from different sources, a source-weighted loss is used:

[0049] Where j is the index of the data sample. Let be the predicted effective wave height of the control point in front of the dike in the j-th sample, as output by the model. Let j be the effective wave height value of the control point in front of the dike measured in the field in the j-th sample. Let J be the effective wave height value of the control point in front of the dike obtained by SWAN simulation in the j-th sample. This represents the number of samples measured on-site. The number of simulated samples for SWAN. These are the on-site measured and SWAN data loss weighting coefficients, respectively. This ensures that the field observation data in front of the seawall has a calibration effect on the model output no less than that of the SWAN auxiliary samples. The field measurement data is only used for training and validating the wave height reduction performance in front of the seawall; the SWAN simulation results are used to supplement the open sea wave condition combinations not covered by the field measurements.

[0050] (e) Overall loss and gradient alignment adaptive weight adjustment: Overall loss of the neural network conversion model for space physical information Data loss Physical loss and boundary loss Three-weighted composition:

[0051] in, For the weighting coefficients of the data loss term, The weighting coefficient for the physical loss term. These are the weight coefficients for the boundary loss term (fixed weights, not involved in adaptive adjustment).

[0052] In a preferred embodiment of the present invention, a gradient alignment adaptive weight adjustment strategy is adopted to dynamically adjust the data loss weight and the physical loss weight. When a conflict in the gradient directions of the two is detected, the physical loss weight is reduced and the data loss weight is increased, as detailed below: To avoid training instability caused by conflicting gradient directions among loss terms, an adaptive weight adjustment strategy based on gradient alignment is adopted. The gradient direction consistency coefficient between the physical loss and the data loss is defined. :

[0053] in, and These are the gradient vectors of the physical loss and data loss with respect to the network parameters, respectively.

[0054] When detected When this occurs, it indicates a conflict in the gradient directions of the two. To maintain the relative balance between physical loss and data loss in the total gradient, the corresponding weights are updated with boundary constraints according to the following rules ( (Index for training iteration steps)

[0055] in, To preset the adjustment step size, the value of α is determined based on the balance between model training convergence speed and stability, with a preferred range of 0.01 to 0.1. This range has been experimentally verified to balance convergence speed and training stability in wave calculation scenarios. Those skilled in the art can adjust it within this range based on the loss curve observed during actual training. This is a truncation function. and To preset the upper and lower limits of the weights, For the first The physical loss weighting coefficient of the step. For the first Temporary weights for physical loss after truncation. Indicates the number of training iterations. Indicates physical loss (physics). The temporary weights for the data loss after truncation in the (t+1)th training iteration are given. Let be the data loss weight coefficient for the t-th training iteration. This indicates data loss.

[0056] The adjusted temporary weights are updated using a scaling normalization method that keeps the sum of the two weights constant (fixed sum of 2):

[0057] This ensures that when gradient directions conflict, the physical loss weights are dynamically reduced and the data loss weights are correspondingly increased, preventing the network fitting direction from being skewed due to excessive physical residuals.

[0058] The effective wave height in front of the dike output in step S2 has high conversion accuracy, which effectively ensures the accuracy of subsequent power generation status identification and power generation duration statistics.

[0059] S3: Based on the preset power generation threshold wave height of at least two wave energy generator units, determine hourly whether the incoming waves from the ocean cause each unit to be in an effective power generation state, and take the effective power generation state of any one unit as the effective power generation state of the power station to obtain the hourly effective power generation state sequence corresponding to the incoming waves from the ocean. Step S3 is used to determine the effective power generation status based on the wave direction interval path and the unit threshold.

[0060] like Figure 3 The diagram shown is a logical schematic of the hourly union correction of incoming waves from the open sea and wind-generated waves within the harbor, according to an embodiment of the present invention. Ten years of hourly ERA5 data are input into the trained PINN model. Based on the dominant wave direction in the open sea, the corresponding wave direction interval and propagation path are determined. The spatial coordinates, water depth function, and dissipation parameters of this propagation path are substituted into the PINN conversion model to output the time series of effective wave heights in front of the breakwater. The offshore wave direction-dikeway wave direction interval mapping table is used to determine the corresponding propagation path number, along-path depth function, along-path dissipation parameter set, and dikeway dominant wave direction interval based on the offshore dominant wave direction.

[0061] The method for determining whether incoming waves from the open sea cause each unit to be in an effective power generation state on an hourly basis, based on the preset power generation threshold wave height of at least two wave energy generator units, is as follows: The hourly power generation state of the two units is determined using the power generation threshold wave height of the first unit and the power generation threshold wave height of the second unit, respectively. The overall effective power generation state of the power station is calculated as a logical OR of the power generation state of each unit. The effective power generation hours for each wave direction interval and each month are counted, and the total effective power generation hours are divided by 24 to calculate the equivalent effective power generation days. Details are as follows: Based on on-site measurements and calibration, the wave period in the sea area of ​​this project is concentrated within the effective response range of the device, and the impact of wave direction changes on the start-up threshold is negligible. Therefore, a fixed threshold is adopted in this embodiment. For the different start-up characteristics of 50kW and 100kW units, hourly power generation states are defined respectively. , (These are the fixed levee frontage power generation threshold wave heights for 50kW and 100kW units, respectively).

[0062] In this embodiment of the invention, effective power generation of a power station is defined as at least one generating unit having power output, then:

[0063] in, All are hourly power generation status indicator variables (1 for power generation, 0 for no power generation). For a 50kW unit The hourly power generation status indicator variable at any given time. For a 100kW unit The hourly power generation status indicator variable at any given time. The wave is an hourly comprehensive effective power generation status indicator variable for the power station caused by waves coming from the open sea. The subscript wave indicates the waves coming from the open sea. A value of 1 indicates that at least one unit is generating electricity, and a value of 0 indicates that neither unit is generating electricity. : Indicator function, which takes the value 1 when the logical condition is true, and takes the value 0 otherwise; Spatial PINN model predictions Effective wave height at the control point in front of the dike at any given time, subscript Indicates significant wave height, subscript Indicates the area in front of the dike (inner), the independent variable. Indicates time; The maximum value operator is used to find the maximum value among multiple elements in a collection. Indicates the time.

[0064] The effective power generation hours for each wave direction interval and each month are calculated, and the total effective power generation hours are divided by 24 to convert them into equivalent effective power generation days. For application scenarios where the period or wave direction has a significant impact, the above fixed threshold can be extended to a function that varies with wave direction and period. .

[0065] S4: Based on the wind-forming wave power generation proxy criterion composed of wind speed, wind direction and continuous duration, the wind-forming wave power generation status in the port is determined hourly, and the time period corresponding to the wind-forming wave power generation status in the port is logically ORed with the hourly effective power generation status sequence corresponding to the incoming waves in the open sea to obtain the comprehensive effective power generation status sequence of the power station. Step S4 is used to achieve hourly union correction of incoming waves from the open sea and wind-generated waves in the harbor.

[0066] In the narrow waters within the harbor, localized wind-generated waves can form under NNW wind conditions. To avoid double counting of waves from the open sea and wind-generated waves, a time-by-time logical union operation is employed.

[0067] In the criterion for wind-induced wave generation, the wind speed threshold Wind direction range and duration threshold These parameters are calibrated based on the effective wind distance of the port area, on-site wind condition records, and device operation records. When determining the wind-induced wave power generation state within the port hourly, if the hourly wind speed reaches the aforementioned wind speed threshold, the hourly wind direction falls within the aforementioned wind direction range, and the continuous duration reaches the aforementioned duration threshold, the corresponding time is determined to be the wind-induced wave power generation state within the port. Specifically: Hourly wind speeds obtained using ERA5 or on-site observations ,wind direction and continuous duration Define the state of wind-induced wave power generation. :

[0068] in, This is an hourly power generation status indicator variable for wind-driven waves.

[0069] When the aforementioned proxy criterion is met, the corresponding time will be identified as a power generation time that may be triggered by wind-induced waves within the harbor.

[0070] The comprehensive effective power generation state sequence of the power station is obtained by performing an hourly logical OR operation on the hourly effective power generation state sequence corresponding to the incoming waves in the open sea and the time period corresponding to the wind-induced wave power generation state in the harbor.

[0071] in, This is an indicator variable for the comprehensive and effective power generation state sequence.

[0072] S5: Calculate the total equivalent number of power generation days for the whole year based on the comprehensive effective power generation status sequence of the power station; In this embodiment of the invention, step S5 is used to calculate the comprehensive equivalent number of power generation days for the whole year.

[0073] Annual equivalent number of power generation days Based on the power plant's comprehensive effective power generation state sequence The overlapping time periods are obtained by summing the data hourly and dividing by 24. The overlapping time periods are directly determined by the hourly state sequence.

[0074] Where year represents the calendar year (e.g., 2015,...,2024). This represents the total equivalent number of days of power generation throughout the year. The overlapping time periods are directly determined by the hourly state sequence, rather than indirectly calculated through two separate statistical values.

[0075] In a preferred embodiment, step S5 of this invention is further used to calculate the monthly power generation. Specifically, based on the effective wave height time series of the control point in front of the dike and the wave-to-electricity conversion power characteristics of the wave energy power generation device, the hourly electrical power is calculated, and the hourly electrical power is multiplied by the time step and then accumulated to obtain the monthly power generation. The wave-to-electricity conversion power characteristics are used to characterize the efficiency relationship of the wave energy power generation device in converting wave energy into electrical energy. Its specific form includes, but is not limited to, a wave height-period-power interpolation function, or a combination of a wave-to-electricity comprehensive conversion efficiency function and an equivalent capture width.

[0076] The hourly power is calculated using any two of the following methods: Method 1: The hourly output of each generating unit is determined based on the wave height-period-power interpolation function, using the effective wave height and average period in front of the dike. The output is then limited by the rated power of each unit. The instantaneous total output of the power station is the sum of the outputs of all generating units, and is further limited by the upper limit of the grid-connected feeder capacity. Details are as follows: Based on the wave height sequence in front of the dike and the performance characteristics of the device, hourly electrical power is calculated. The influence of wave direction has been reflected through the zoned propagation path from the open sea to the dike and the wave height conversion process in front of the dike, and is not repeated in the single-unit power curve.

[0077] In embodiments of this invention, a two-dimensional wave height-period-power relationship table is used to construct a single-unit power interpolation function. The power curve can be established using any one of the following: equipment factory performance curve, model test results, or field commissioning data. The single-unit output of 50 kW and 100 kW units is calculated respectively:

[0078] in, These are the wave height-period-power interpolation functions for the two generating units, respectively. : Hourly power of a 50 kW unit after being limited by the rated power of a single unit; the subscript 50 indicates a 50 kW unit. Hourly power of a 100kW unit after being limited by the rated power of a single unit; the subscript 100 indicates a 100kW unit. Indicates time; Predictions based on a friction wave height reduction model using spatial PINN. Effective wave height at the control point in front of the dike at any given time, in meters, subscript Indicates significant wave height, subscript Indicates the area in front of the dike. The average period of time, expressed in seconds; : The rated power of a single unit is taken as 50 kW; : The rated power of a single unit is taken as 100 kW; Minimum value operator.

[0079] Method 2: The wave energy capture power of each generator unit is calculated based on its wave-to-electricity integrated conversion efficiency function η(·) and equivalent capture width, combined with the effective wave height and wave group velocity in front of the dike. The power is then limited by the rated power of each unit. The instantaneous total output of the power station is the sum of the outputs of all units, and the overall power is limited by the upper limit of the grid-connected feeder capacity. Details are as follows: As an alternative, if the device has an equivalent capture width Then it can be calculated as follows:

[0080] in, : Unit wave-to-electricity integrated conversion efficiency function; : Unit wave-to-electricity integrated conversion efficiency function; : The equivalent capture width of the unit, in meters, represents the incident wavefront width through which the device effectively absorbs wave energy. : The equivalent capture width of the unit, in meters, represents the incident wavefront width through which the device effectively absorbs wave energy. : Wave group velocity at time t, in units of The value is calculated from the water depth in front of the dike and the average period according to the linear wave dispersion relation.

[0081] The instantaneous total output of the power station is the sum of the outputs of the two generating units, and is subject to the overall limit of the grid-connected feeder capacity:

[0082] in, The instantaneous total output of the power station is the sum of the outputs of the two generating units, and the unit is kW; Hourly power output of the power station after the overall limitation of the grid-connected feeder capacity, in kW; The upper limit of the grid-connected feeder capacity of the power station is expressed in kW. In this embodiment, the preferred value is 150 kW, which is the sum of the rated installed capacity of the two units.

[0083] After calculating the hourly power using either Method 1 or Method 2, multiply the hourly power after overall limiting by the time step, and sum them up to obtain the monthly power generation. for:

[0084] in, Indexed by month, Time step (hourly data) ), For the first The cumulative power generation over the month.

[0085] Example 2: This embodiment uses a local OWC wave energy fusion demonstration project as an example to provide a detailed description of the method of the present invention.

[0086] (I) Project Overview: A 264.5m long and 54.5m wide breakwater is planned within the port terminal area of ​​a certain location. The project will modify two caissons at the end of the breakwater into 50kW and 100kW OWC units respectively – a caisson terminal integrated system. The wave height at startup of the 50kW unit is approximately 0.4m, and that of the 100kW unit is approximately 0.5m.

[0087] (II) ERA5 Data Extraction: Using a reference point approximately 24 km northwest of the project site (121°E, 39°N) in the open sea, hourly significant wave height, mean period, and dominant wave direction were extracted from ERA5 reanalysis data (2015–2024, a total of 10 years). The ERA5 data has a spatial resolution of approximately 0.25° × 0.25° and a temporal resolution of 1 hour. After extraction, the coordinates were matched with the project coordinates, and linear interpolation was used to process missing values.

[0088] (III) Construction and training of the spatial PINN wave height calculation model: (a) Wave direction interval and propagation path settings: Propagation paths were established for four offshore wave direction intervals: WNW, W, WSW, and SW. Each path extended from the offshore reference point to the control point in front of the breakwater, and the corresponding water depth profile was extracted based on engineering nautical charts and SWAN topographic data. The length of each path was determined based on the actual propagation path; in this embodiment, the lengths of the four paths were approximately [missing information]. Between these points, a continuous and first-order differentiable water depth function is constructed using cubic spline interpolation for the discrete water depth node data extracted from each propagation path. This is used for subsequent physical residual calculations.

[0089] (b) Network structure: Network structure: A 6-layer fully connected neural network, with 64 neurons per layer, and the activation function is tanh. The main network input is... The output is the wave height of each node along the current path. . water depth The known background field is introduced into the physical residual calculation after cubic spline interpolation and is not included in the network input.

[0090] (c) Physical constraints: Constructed according to the wave energy flow residual formula in Example 1 Each training batch samples physical configuration points from four propagation paths, and the total number of physical configuration points across the four paths is... The number of physical points along each path is proportionally allocated based on the path length. When calculating physical loss, the path arc length coordinates are used. Set as a differentiable variable; PINN output The water depth function obtained by differentiable interpolation They jointly participate in the calculation of group velocity and wave energy flow, and calculate the spatial derivative through automatic differentiation. The bottom friction dissipation parameters and breaking dissipation parameters adopt the corresponding parameter forms of the publicly available SWAN parameterized model, which can be determined according to the bottom sediment type and wave breaking characteristics of the target sea area.

[0091] (d) Boundary constraints and data constraints: Boundary constraints: at the offshore reference point constraint .

[0092] Data constraints: Data loss is jointly caused by two months of field-measured wave height data in front of the breakwater (sampling frequency 1Hz, processed into hourly significant wave heights) and SWAN simulation results (covering 30 offshore wave condition combinations). Field-measured data were divided into a calibration set (first 6 weeks) and a validation set (last 2 weeks) according to continuous time periods, without random shuffling to avoid data leakage between adjacent time periods. In the source-weighted calculation... The on-site measured single sample weight is no less than the SWAN sample weight.

[0093] (e) Training: Initial learning rate The gradient decays every 500 steps. Each batch contains 1800 physical configuration points and 248 data samples (a 1:3 mix of field and SWAN samples), for a total batch size of 2048. A total of 50,000 iterations are performed. The gradient direction consistency coefficient is calculated every 1000 steps. Adjust according to the update rules with boundary constraints. and .

[0094] Model validation: The model performance was evaluated using field-measured data from the validation set that were not used in the training. The root mean square error (RMSE), mean absolute error (MAE), and bias of the predicted wave height in front of the dike were calculated, and the classification consistency was statistically analyzed separately near the threshold (0.3m to 0.7m).

[0095] (iv) Effective power generation status determination and days statistics: Hourly ERA5 data from 2015 to 2024 were input into the PINN model. Based on the dominant wave direction in the open sea, the corresponding wave direction intervals and propagation paths were determined, and the seawall wave height sequence was output. The mapping table of open sea wave direction to seawall wave direction intervals is shown in Table 1. Table 1

[0096] respectively To determine the hourly power generation status of the two generating units, the effective power generation of the power station is defined as follows: The effective power generation hours for each year from 2015 to 2024 are statistically analyzed, and divided by 24 to obtain the equivalent effective power generation days for each year. This embodiment uses the spatial PINN model and wave energy density coefficient... After calculation using the aforementioned fixed thresholds, the ten-year average annual equivalent effective power generation days from offshore waves is 81.3 days, with the annual values ​​ranging from approximately... sky.

[0097] (V) Wind forms waves and gathers and corrects over time: Based on field observations and ERA5 wind speed data, the NNW wind direction range (337.5°±22.5°, i.e., 315°-360°, wind speed threshold) was determined. (Corresponding to level 4 wind), duration threshold Hourly judgment According to ten years of statistics, the equivalent effective power generation days for wind-induced waves is 13.8 days.

[0098] Will and Hourly logical OR operations yielded an average annual equivalent effective power generation day of 89.8 days over ten years. This value is less than the arithmetic sum of 95.1 days (81.3 days for offshore waves and 13.8 days for wind-induced waves); the difference is 5.3 days, corresponding to approximately 127.2 hours of overlapping time periods removed in the hourly logical union operation, accounting for approximately 1.45% of the total annual hours. These overlapping time periods are directly determined by the hourly state sequence.

[0099] (vi) Monthly power generation calculation: In this second embodiment, a two-dimensional wave height-period-power relationship table is used to construct a single-unit power interpolation function: based on the wave height-period-power relationship table measured in model tests and on-site commissioning of 50kW and 100kW units, a two-dimensional interpolation function is constructed. and The influence of wave direction has been reflected through the zoned propagation path from the open sea to the seawall and the wave height calculation process in front of the seawall, and is not repeated in the single-unit power curve. The total instantaneous power output of the power station is obtained by summing the power outputs after limiting the rated power of each unit, and then by overall limiting the grid-connected feeder capacity (150kW). The monthly power generation is obtained by multiplying the hourly power output by 1 hour and summing them. The results show that power generation begins to increase in October each year, reaches a relatively high level from November to February of the following year, which is consistent with the analysis of the number of days of power generation; power generation is relatively low from April to August.

[0100] Furthermore, embodiments of the present invention also include a control device. In one embodiment of the control device according to the present invention, the control device includes a processor and a storage device. The storage device can be configured to store a program for executing the method of calculating the annual power generation duration of an OWC wave energy power station based on ERA5 data and wave height conversion as described in the above-described method embodiments. The processor can be configured to execute the program stored in the storage device, which includes, but is not limited to, the program for executing the method of calculating the annual power generation duration of an OWC wave energy power station based on ERA5 data and wave height conversion as described in the above-described method embodiments. For ease of explanation, only the parts related to the embodiments of the present invention are shown. For specific technical details not disclosed, please refer to the method section of the embodiments of the present invention. The control device can be a control device device device comprising various electronic devices.

[0101] Furthermore, embodiments of the present invention also include a computer-readable storage medium. In one embodiment of the present invention, the computer-readable storage medium may be configured to store a program that executes the method for calculating the annual power generation duration of an OWC wave energy power station based on ERA5 data and wave height conversion as described in the above-described method embodiments. This program may be loaded and run by a processor to implement the above-described method for calculating the annual power generation duration of an OWC wave energy power station based on ERA5 data and wave height conversion. For ease of explanation, only the parts related to the embodiments of the present invention are shown; for specific technical details not disclosed, please refer to the method section of the embodiments of the present invention. The computer-readable storage medium may be a storage device comprising various electronic devices. Optionally, in the embodiments of the present invention, the computer-readable storage medium is a non-transitory computer-readable storage medium.

[0102] Furthermore, it should be understood that since the various modules are only provided to illustrate the functional units of the device of the present invention, the physical devices corresponding to these modules may be the processor itself, or a part of the processor's software, hardware, or a combination of software and hardware. Therefore, the number of modules shown in the figures is merely illustrative.

[0103] Those skilled in the art will understand that the various modules in the device can be adaptively split or combined. Such splitting or combining of specific modules will not cause the technical solution to deviate from the principles of the present invention; therefore, the technical solutions after splitting or combining will fall within the protection scope of the present invention.

[0104] The technical solution of the present invention has been described above with reference to the preferred embodiments shown in the accompanying drawings. However, it will be readily understood by those skilled in the art that the scope of protection of the present invention is obviously not limited to these specific embodiments. Without departing from the principles of the present invention, those skilled in the art can make equivalent changes or substitutions to the relevant technical features, and the technical solutions after these changes or substitutions will all fall within the scope of protection of the present invention.

Claims

1. A method for estimating the annual power generation duration of an OWC wave energy power station based on ERA5 data and wave height conversion, characterized in that, Includes the following steps: Hourly wave element data of offshore reference points in the engineering area are extracted from the ERA5 global reanalysis dataset. The wave element data includes significant wave height, mean period, and main wave direction. The hourly wave element data is input into a pre-constructed and trained spatial physical information neural network conversion model along the wave propagation path, and the effective wave height time series of the control points in front of the dike is output. Based on the preset power generation threshold wave height of at least two wave energy generator units, it is determined hourly whether the incoming waves from the ocean cause each unit to be in an effective power generation state, and the effective power generation state of any one unit is taken as the effective power generation state of the power station, thus obtaining the hourly effective power generation state sequence corresponding to the incoming waves from the ocean. Based on the wind-forming wave power generation proxy criteria consisting of wind speed, wind direction and continuous duration, the wind-forming wave power generation status in the port is determined hourly. The time period corresponding to the wind-forming wave power generation status in the port is then logically ORed with the hourly effective power generation status sequence corresponding to the incoming waves from the open sea to obtain the comprehensive effective power generation status sequence of the power station. The annual equivalent number of power generation days is calculated based on the comprehensive effective power generation status sequence of the power station.

2. The method for estimating the annual power generation duration of an OWC wave energy power station based on ERA5 data and wave height conversion according to claim 1, characterized in that, The construction method of the space physical information neural network conversion model includes: Divide the wave direction into at least two wave direction intervals based on the incident wave direction in the open sea. For each wave direction interval, a corresponding propagation path is preset between the reference point in the open sea and the control point in front of the dike. The propagation path is discretized into multiple path nodes along its arc length coordinate. The spatial physical information neural network conversion model takes the propagation path arc length coordinates, the effective wave height of the offshore reference point, the average period of the offshore reference point, and the sine and cosine values ​​of the main wave direction of the offshore reference point as inputs, and the effective wave height of each node along the current propagation path as output. The node at the end of the propagation path corresponds to the control point in front of the dike, and the effective wave height at that node is the effective wave height of the control point in front of the dike.

3. The method for estimating the annual power generation duration of an OWC wave energy power station based on ERA5 data and wave height conversion according to claim 2, characterized in that, The spatial physical information neural network conversion model is a multi-layer fully connected neural network. Its input layer receives the arc length coordinates of the propagation path, the effective wave height of the offshore reference point, the average period of the offshore reference point, and the sine and cosine values ​​of the main wave direction of the offshore reference point. After processing by multiple fully connected hidden layers, the output layer outputs the effective wave height of each node along the current propagation path.

4. The method for estimating the annual power generation duration of an OWC wave energy power station based on ERA5 data and wave height conversion according to claim 2, characterized in that, The training process of the space physics information neural network conversion model embeds the physical constraint of conservation of wave energy flow along the propagation path. The physical constraint is constructed based on the physical law that balances the change of wave energy flow along the propagation path with the dissipation term. The dissipation term includes bottom friction dissipation rate and wave breaking dissipation rate. The physical loss is the weighted summation of the residuals of wave energy flow along the propagation path at each physical configuration point.

5. The method for estimating the annual power generation duration of an OWC wave energy power station based on ERA5 data and wave height conversion according to claim 2, characterized in that, The overall loss of the space physical information neural network conversion model is composed of a weighted average of data loss, physical loss, and boundary loss. The data loss is used to constrain the consistency between the model output and the measured data and numerical simulation results. The physical loss is used to constrain the model output to satisfy the conservation of wave energy flow along the path. The boundary loss is used to constrain the model output at the offshore reference point to be consistent with the input wave height. Furthermore, a gradient alignment adaptive weight adjustment strategy is adopted to dynamically adjust the weights of the data loss and physical loss. When a conflict in the gradient directions of the two is detected, the weight of the physical loss is reduced and the weight of the data loss is increased.

6. The method for estimating the annual power generation duration of an OWC wave energy power station based on ERA5 data and wave height conversion according to claim 2, characterized in that, The method for determining hourly whether incoming waves from the open sea cause each unit to be in an effective power generation state based on the preset power generation threshold wave height of at least two wave energy generator units is as follows: The hourly power generation status of the two units is determined by the wave height of the first unit's power generation threshold and the wave height of the second unit's power generation threshold, respectively. The comprehensive effective power generation status of the waves coming from the sea outside the power station is the logical OR of the power generation status of each unit. The effective power generation hours for each wave direction interval and each month are counted, and the total effective power generation hours are divided by 24 to convert them into equivalent effective power generation days.

7. The method for estimating the annual power generation duration of an OWC wave energy power station based on ERA5 data and wave height conversion according to claim 1, characterized in that, In the aforementioned wind-wave power generation proxy criterion, the wind speed threshold, wind direction range, and duration threshold are calibrated based on the effective wind distance of the port area, on-site wind condition records, and device operation records. When determining the wind-induced wave power generation status within the harbor on an hourly basis, the corresponding time is determined to be the wind-induced wave power generation status within the harbor when the hourly wind speed reaches the wind speed threshold, the hourly wind direction falls within the wind direction range, and the continuous duration reaches the duration threshold.

8. The method for estimating the annual power generation duration of an OWC wave energy power station based on ERA5 data and wave height conversion according to claim 1, characterized in that, The comprehensive effective power generation state sequence of the power station is an hourly state sequence obtained by performing an hourly logical OR operation on the hourly effective power generation state sequence corresponding to the wave-forming power generation state in the open sea and the time period corresponding to the wind-forming wave power generation state in the harbor. The annual comprehensive equivalent power generation days are obtained by summing the comprehensive effective power generation state sequence of the power station hourly and then dividing by 24. The overlapping periods are directly determined by the hourly state sequence.

9. The method for estimating the annual power generation duration of an OWC wave energy power station based on ERA5 data and wave height conversion according to claim 1, characterized in that, Also includes: Based on the effective wave height time series of the control point in front of the dike and the wave-to-electric power conversion characteristics of the wave energy generation device, the hourly electrical power is calculated. The hourly electrical power is multiplied by the time step and then summed to obtain the monthly power generation. The hourly electrical power is calculated in the following way: Based on the wave height-period-power interpolation function of each unit, the hourly output of each unit is determined by the effective wave height and average period in front of the dike, and the amplitude is limited by the rated power of each unit. The instantaneous total output of the power station is the sum of the outputs of each unit, and the overall output is limited by the upper limit of the grid-connected feeder capacity of the power station. The hourly power output after overall limiting is multiplied by the time step, and then summed to obtain the monthly power generation.

10. The method for estimating the annual power generation duration of an OWC wave energy power station based on ERA5 data and wave height conversion according to claim 1, characterized in that, Also includes: Based on the effective wave height time series of the control point in front of the dike and the wave-to-electric power conversion characteristics of the wave energy generation device, the hourly electrical power is calculated. The hourly electrical power is multiplied by the time step and then summed to obtain the monthly power generation. The hourly electrical power is calculated in the following way: Based on the wave-electric integrated conversion efficiency function and equivalent capture width of each unit, combined with the effective wave height in front of the dike and the wave group velocity, the wave energy capture power of each unit is calculated, and the amplitude is limited by the rated power of each unit. The instantaneous total output of the power station is the sum of the outputs of each unit, and the overall output is limited by the upper limit of the grid-connected feeder capacity of the power station; The hourly power output after overall limiting is multiplied by the time step, and then summed to obtain the monthly power generation.

Images