Energy Storage Power Station Intelligent Management System

Abstract

The present application discloses an intelligent management system for an energy storage power station, belonging to the technical field of battery management. An intelligent management system for an energy storage power station includes a photovoltaic monitoring module, a wind energy monitoring module, an energy storage monitoring module, a grid connection control module, a power generation prediction module, a power quantity prediction module, and a battery function control module. Based on the historical power generation data of the next time period, the time period of connecting to the power grid, and the output power of electricity, the charging time period and discharging time period of each battery are generated, and the cyclic charging and discharging of each battery are controlled. In the technical solution provided by the present application, the historical power generation data of the next time period is accurately obtained first, so the electric energy that needs to be charged into the energy storage module in a future period of time can be obtained. At the same time, the output power of electricity in the next time period is preset in advance, avoiding the situation that the battery urgently needs to be charged and discharged, and increasing the service life of the battery.

Description

Technical Field

[0001] The present application relates to the technical field of batteries, and more particularly, to an intelligent management system for an energy storage power station. Background Art

[0002] A microgrid is a small-scale power generation, distribution, and consumption system composed of distributed power sources (such as solar energy, wind energy, energy storage systems, etc.), electrical loads, power distribution facilities, monitoring and protection devices, etc. It can achieve internal power balance and can exchange energy with the external power grid.

[0003] A microgrid is generally established with an energy storage power station as the core. The energy storage power station includes a wind power module, a photovoltaic module, an energy storage module, and a grid connection management module. Among them, the wind power module is used for wind power generation, the photovoltaic module is used for solar power generation, and the energy storage module is used for storing electric energy. The grid connection management module is used to manage when the microgrid accesses the power grid.

[0004] In order to ensure the highest efficiency, the microgrid generally stores electricity at night and outputs electricity during the day. In order to ensure the power quality of the microgrid, it is necessary to ensure that the electricity output by the microgrid during the day is stable at a preset value. For this reason, in the existing solutions, based on the historical power generation data of wind and solar power, the average power that the microgrid can output every day is calculated, and then the electric energy stored in the energy storage module is input into the power grid at the rated power during the peak period of the electricity price every day.

[0005] However, in actual operation, the time when the microgrid transmits electricity to the power grid is generally from 2 pm to 4 pm. This period is the period when the power is most in short supply, the period when the solar energy is most abundant, and the period when the photoelectric conversion efficiency of the photovoltaic module in the microgrid is the highest. In this way, the energy storage module needs to be charged and discharged at the same time, which will cause interference to the chemical reaction inside the battery in the energy storage module, resulting in a decrease in battery efficiency and a reduction in battery life. Summary of the Invention

[0006] This section of the present application is used to briefly introduce the concepts, which will be described in detail in the following detailed implementation section. This section of the present application is not intended to identify the key features or essential features of the claimed technical solution, nor is it intended to limit the scope of the claimed technical solution.

[0007] In order to solve the technical problems mentioned in the above background art section, the present application provides an intelligent management system for an energy storage power station, including:

[0008] A photovoltaic monitoring module, configured to obtain the photoelectric generation data of the photovoltaic power generation module;

[0009] A wind energy monitoring module, configured to obtain the wind power generation data of the wind power generation module;

[0010] Energy storage monitoring module, used to detect the power consumption data of each battery in the energy storage module;

[0011] Grid connection control module, used to pre-enter the time period when the microgrid accesses the power grid and the output power;

[0012] Power generation prediction module, calculates the historical power generation data for the next time period based on the photovoltaic generation data and the wind-solar power generation data;

[0013] Battery power prediction module, generates the power of each battery based on the power consumption data to obtain the charge and discharge capabilities of each battery;

[0014] Battery function control module, based on the historical power generation data for the next time period, the time period of accessing the power grid, and the output power, generates the charging time period and the discharging time period of each battery, and controls each battery to cycle charge and discharge.

[0015] In the technical solution provided by the present application, first, the historical power generation data for the next time period is accurately obtained, so the electric energy that needs to be charged into the energy storage module within a future period of time can be obtained. At the same time, the output power for the next time period is preset, so the electric energy that needs to be discharged within a future period of time is determined. In this way, the battery function control module allows the battery to cycle charge and discharge based on this. Furthermore, for each individual battery, it only needs to perform the charging operation or the discharging operation. Therefore, it is avoided that the battery needs to be charged and discharged at the same time, increasing the service life of the battery.

[0016] During the charging and discharging process of the battery, it is necessary to try to keep the charging current and the discharging current of a single battery within a reasonable range to avoid the charging current being too large or too small. Since the battery pack in the energy storage module needs to be divided into a charging group and a discharging group, this will inevitably lead to a situation where if the proportion of the number of batteries in the charging group and the discharging group is not reasonably arranged during charging and discharging of the energy storage module, the charging current or the discharging current will be too high. Based on this, the present application provides the following technical solution:

[0017] Battery function control module, calculates the number of batteries that need to be charged in the future based on the historical power generation data for the next time period, and calculates the maximum output power for the future time period based on the number of batteries that need to be charged in the future.

[0018] In the technical solution provided by the present application, considering the uncontrollability of charging, first, the number of batteries that need to be charged is calculated based on the historical power generation data. After determining the number of charging batteries, the number of batteries that can be discharged is then determined. In this way, based on the number of discharging batteries, the maximum output power is finally determined. In this way, a reasonable output power can be set when accessing the power grid, thereby avoiding too large a charging current or a discharging current during the cyclic operation of the battery, which affects the service life.

[0019] Generally speaking, the randomness of wind power is much greater than that of photovoltaic power. If wind power and photovoltaic power are directly used with the same model, the model will tend to overfit during the training of big data, resulting in low model accuracy. To address this issue, the present application provides the following technical solutions:

[0020] The power generation prediction module includes:

[0021] A wind power prediction unit for predicting the power generation data of wind power in the next time period;

[0022] A photovoltaic power prediction unit for predicting the power generation data of photovoltaic power in the next time period.

[0023] In the technical solutions provided by the present application, wind power and photovoltaic power are predicted using two different units, so the accuracy of the total power generation prediction can be ensured.

[0024] The battery's power is predicted based on the current charge capacity of the battery, that is, the actual charge content in the battery is calculated according to the charging current or discharging current of the battery. Although the charge entering and leaving the battery can be monitored by current. However, in fact, the amount of charge that the battery can charge (power) and the amount of charge output are related to the predicted life of the battery. The better the battery life (the lower the internal resistance), the lower the energy loss during the charging and discharging processes of the battery, and vice versa. Therefore, when predicting the maximum charge and maximum discharge (discharge capacity) of the battery, it is necessary to accurately evaluate the performance (predicted life) of the battery, and the evaluation of the battery life is a relatively dynamic process. Simply based on the number of times the battery is used and its usage status, the predicted life of the battery cannot be accurately evaluated. In the traditional use of battery packs, since multiple batteries are managed as a whole for charging and discharging, big data (the normal distribution characteristics of battery performance) can be used to accurately predict the discharge capacity and charging capacity of a battery pack. However, in this solution, for each battery, the discharge group and the charging group need to be replaced, so it is necessary to accurately predict the performance of a single battery. Therefore, to ensure the accuracy of battery performance prediction, the present application provides the following technical solutions:

[0025] Further, the power prediction module includes:

[0026] A data collection unit for collecting the power consumption data of each battery i, where i is the index of the battery;

[0027] A data preprocessing unit for preprocessing the power consumption data of each battery i to obtain the prediction data H of each battery i ;

[0028] A data prediction unit, and the prediction data of each battery is input into a CNN neural network model to obtain the predicted life of each battery;

[0029] A charge-discharge performance calculation unit, which converts the predicted life into a corresponding conversion ratio, calculates the current standard charge-discharge amount according to the charge-discharge data, and multiplies the conversion ratio by the standard charge-discharge amount to calculate the charge-discharge capacity of each battery.

[0030] In the technical solution provided by this application, data collection, data preprocessing are performed for each battery, and corresponding prediction data is generated. Then the prediction is input into the CNN neural network model, so the predicted life of each battery can be accurately predicted. Then, based on the predicted life, the charge-discharge capacity of each battery is measured, so the accuracy is higher. In this way, in battery management, it can reach the bottom-level batteries, which is convenient for realizing the charge-discharge management of the batteries.

[0031] Further, the power consumption data includes: charge-discharge voltage vs. time curve, charge-discharge current vs. time curve, charge-discharge temperature vs. time curve.

[0032] Among the above data, the performance change of the battery after multiple cycles of use can be accurately described, thereby increasing the accuracy of model prediction.

[0033] Because this solution needs to accurately predict the performance of each battery, it is necessary to increase the model accuracy. However, increasing the model accuracy will inevitably introduce more input data. If there is too much input data, there will be too many data items, a large amount of redundancy, and the relationship between the weight items inside the model will be complex, making it impossible to accurately analyze the law behind the data. To solve this problem, this application provides the following solution:

[0034] Further, the data preprocessing unit includes:

[0035] A data dimensionality reducer, which is used to reduce the dimension of the prediction data H i for dimensionality reduction;

[0036] A normalizer, which is used to normalize the data;

[0037] A correlation predictor, which is used to perform charge-discharge correlation screening on the data and screen out the data error prediction data.

[0038] In this solution, through data dimensionality reduction, normalization, and screening, the complexity of the data can be reduced, the redundancy of the data can be reduced, thereby reducing the model burden and increasing the running ability and model accuracy of the model.

[0039] Further, the dimension-reduced prediction data H i ={[t min(v1) ,v1],[tmin(v2) , v2], [t min(a1) , A1], [t min(a2) , A2], [t T1 , T1], [t T2 , T2]};

[0040] [t min(v1) , v1] = {SV1 s ≥V max}, s = 1, 2, 3…, t min(v1) is the time when the voltage of the battery first reaches the maximum value during charging, v1 is the voltage value when the battery terminal voltage is first not less than the maximum voltage value, s is the number of charging times, SV1 s is the voltage data of the charging voltage - time curve, V max is the rated maximum charging voltage;

[0041] [t min(v2) , v2] = {SV2 d ≥V min}, d = 1, 2, 3…, t min(v2) is the time when the voltage of the battery first reaches the minimum value during discharging, v2 is the voltage value when the battery terminal voltage is first not greater than the minimum voltage value, d is the number of discharging times; SV2 s is the voltage data of the discharging voltage - time curve, V min is the rated minimum charging voltage;

[0042] [t min(a1) , A1] = {SA2 s ≤A max}, s = 1, 2, 3…, t min(a1) is the time when the current of the battery starts to decline during charging, A1 is the current value when the battery current starts to decline, A max is the maximum current value during charging, s is the number of charging times; SA2 s is the current data of the charging current - time curve;

[0043] [t min(a2) , A2] = {SA2 d ≥Ax1}, d = 1, 2, 3…, t min(a2) is the time when the current of the battery starts to rise during discharging, A2 is the current value when the battery current starts to rise, Ax1 is the maximum current value during discharging, d is the number of discharging times; SA2 d is the current data of the discharging current - time curve;

[0044] [t T1 , T1] = {ST1 s ≥T1 max}, s = 1, 2, 3…, t T1The time when the temperature first reaches the maximum value during battery charging, T1 is the highest temperature reached during battery charging, s is the number of charging times, ST1 s is the temperature data of the charging temperature data and time curve, T1 max is the maximum charging temperature;

[0045] [t T2 , T2] = {ST1 d ≥T2 max}, s = 1, 2, 3…, t T2 The time when the temperature first reaches the maximum value during battery discharging, T2 is the highest temperature reached during battery discharging, d is the number of discharging times, ST2 d is the temperature data of the discharging temperature data and time curve, T2 max is the maximum discharging temperature.

[0046] In the technical solution provided by this application, when performing feature dimensionality reduction, the maximum value, minimum value, and obvious mutation points with the most representative features are found in advance from each data curve. These mutation points are actually the key points of the battery's power consumption data. Therefore, on the basis of effectively reducing the data dimension and reducing data redundancy, data loss is effectively avoided, and further, the accuracy of model prediction is increased.

[0047] When predicting the expected life of a battery, it mainly relies on the collected historical power consumption data of the battery, and the historical power consumption data of the battery. And the collected battery historical data will be affected by the sensor accuracy. After the sensor accuracy decreases, because the collected data itself is incorrect, or the collected data will have a drift without obvious features due to the decrease in the sensor measurement accuracy. In this case, simple data processing cannot detect the change in sensor accuracy, and it can be foreseen that if these data are used to predict the expected life of the battery, it will inevitably lead to a large difference between the prediction result and the actual situation. Based on this, the accuracy will inevitably decrease. For this reason, this application provides the following technical solution:

[0048] Further, for the prediction data, calculate the correlation coefficient P between each data. If the range of the correlation coefficient between each data item is within the preset range, the prediction data is normal. If it is not within the preset range, delete the battery.

[0049]

[0050] Among them, P X,Y represents the correlation coefficient between data item and , and respectively represent the Xth item data and the Yth item data in the prediction data Hi.

[0051] Through the technical solution provided by this application, by calculating the correlation between data, it can be predicted that if there is a large correlation between data, it indicates that the collected data is real, reliable, and valuable data for reference. Only when these data are input into the neural network model for predicting battery life will it have greater value.

[0052] During the process of predicting the expected battery life, the battery will continuously undergo charge and discharge cycles. Therefore, in the generated data, the time series data is very long. So, during the propagation of the neural network, there will be a serious problem of gradient attenuation. Thus, during the operation of the model, it is difficult for the internal parameters of the model to accurately adapt to the characteristics of gradient attenuation and accurately capture the actual characteristics of the change in the expected battery life, thereby affecting the learning ability of the model. Moreover, in practice, the expected battery life (performance) is not a monotonically decreasing process and there will also be an upward trend. For this reason, this application provides the following technical solution:

[0053] Furthermore, the data prediction unit includes:

[0054] A convolutional layer that slides a convolutional kernel on the input prediction data H i to calculate the dot product of the convolutional kernel and the input data, generating a feature map to generate local features of the prediction data H i ;

[0055] A pooling layer that receives the output from the previous convolutional layer as input and reduces the feature dimension;

[0056] An Attention layer that performs weighted summation on the hidden features generated by each convolutional layer to generate the final predicted life score U.

[0057] Furthermore, in the Attention layer,

[0058] R ki = tanh(u k · q ki · u i );

[0059] where R ki represents an intermediate variable, representing the attention weight between the k-th position and the i-th position, u k represents the k hidden states, q ki represents the weight coefficient matrix to be learned, and tanh represents the hyperbolic tangent activation function;

[0060]

[0061] The pki represents the normalized attention weight, indicating the degree of attention of the k-th position to the i-th position;

[0062] R i represents the hidden variable at the k-th position.

[0063] In the technical solution provided by this application, aiming at the technical problems that it is difficult for the internal parameters of the model to accurately adapt to the characteristics of gradient attenuation and it is difficult to accurately capture the actual characteristics of the change in the expected battery life, in the CNN neural network, an Attention layer is introduced, which can filter out the feature data with large weight coefficients and give priority to paying attention to this information. At the same time, the introduced self-attention adaptation mechanism can adaptively allocate weights to the hidden states of each time period to form a more stable long-term dependence relationship, so as to increase the model prediction accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0064] The drawings constituting a part of this application are used to provide a further understanding of this application, making other features, purposes, and advantages of this application more obvious. In addition, throughout the drawings, the same or similar reference numerals represent the same or similar elements. It should be understood that the drawings are schematic, and the elements and elements are not necessarily drawn to scale.

[0065] In the drawings:

[0066] Figure 1 is a schematic structural diagram of the intelligent management system for the energy storage power station.

[0067] Figure 2 is the discharge voltage-time curve under different discharge times.

[0068] Figure 3 is a schematic structural diagram of the data prediction unit.

[0069] Figure 4 is a schematic structural diagram of the Attention layer. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0070] The embodiments of this application will be described in more detail below with reference to the drawings. Although some embodiments of this application are shown in the drawings, it should be understood that this application can be implemented in various forms and should not be construed as limited to the embodiments set forth herein. On the contrary, these embodiments are provided to more thoroughly and completely understand this application. It should be understood that the drawings and embodiments of this application are only for exemplary purposes and are not used to limit the protection scope of this application.

[0071] It should also be noted that, for ease of description, only the parts related to the invention are shown in the drawings. In the absence of conflict, the embodiments and features in the embodiments of the present application can be combined with each other.

[0072] The present application will be described in detail below with reference to the accompanying drawings and in combination with embodiments.

[0073] Reference Figure 1 The intelligent management system of energy storage power station includes photovoltaic monitoring module, wind energy monitoring module, energy storage monitoring module, grid connection control module, operation monitoring module, power generation prediction module, power prediction module and battery function control module.

[0074] The photovoltaic monitoring module is used to obtain the photovoltaic generation data of the photovoltaic power generation module; the photovoltaic generation data here refers to the historical power generation data of the solar panels of the energy storage station, which generally includes the power generation data of the previous period and the power generation data of the current time in previous years. The same is true for wind power generation data. The wind energy monitoring module is used to obtain the wind power generation data of the wind power generation module. The operation monitoring module is used for intelligent alarm of operation status, intelligent charging and discharging management, loss analysis, and power statistics analysis. The photovoltaic monitoring module, wind energy monitoring module and operation monitoring module are all functional modules commonly used in current energy storage stations, and the specific construction method will not be described here.

[0075] The energy storage monitoring module is used to detect the power consumption data of each battery in the energy storage module; each battery here is the smallest energy unit that can be controlled by the energy storage power station. Some batteries are fixed as battery packs through circuits, and it is impossible to switch the charge and discharge of individual batteries. Some batteries can switch the charge and discharge individually for each battery, which is related to the setting position of the sensor and the hardware circuit design. In this solution, in order to increase the accuracy, it is necessary to set a separate charge and discharge switch for each battery. The specific circuit design scheme is not shown here, because the current output by a single lithium battery is not large, so the corresponding circuit design is not complicated, and the circuit structure is not further provided here.

[0076] The grid-connected control module is used to pre-enter the time period for the microgrid to connect to the grid and output the electric power. The purpose of building the energy storage power station is actually to input the electricity generated from new energy into the grid. Different from other energies, electricity cannot be stored in the grid. Therefore, the grid needs to adjust the power structure according to the input and load on the grid. Due to the instability of solar and wind energy, they do not meet the conditions for inputting into the overall grid. For this reason, currently, the clean energy collected is stored in the battery, and a stable power transmission request is sent to the grid in advance during the time when electricity consumption is the most intense every day (afternoon during the day), and then electricity is transmitted to the grid during the corresponding time period. In this way, the disordered clean energy can be transformed into orderly energy. For example, if the energy storage station finds at night that its electric energy can be output at a power of 100 kWh for 2 hours, it will send a request to the State Grid to transmit electricity at a power of 100 kWh for 2 hours from 2 pm to 4 pm.

[0077] Of course, in practice, to ensure efficiency, generally, a request will not be sent after the energy storage station reaches the maximum energy storage efficiency, but it will be dynamically adjusted. For example, if it is calculated that the energy storage station can be fully charged by tomorrow afternoon at 2 pm, the power-on request will be sent to the State Grid in advance, so that clean energy will not be wasted during the power transmission process.

[0078] This part of the dynamic adjustment can achieve the maximum benefit according to the dynamic electricity price of the State Grid under the condition of known power generation. The specific grid-connected management method will not be described here. The power generation prediction module calculates the historical power generation data for the next time period based on the photovoltaic generation data and the wind-solar power generation data; the power generation prediction module includes: a wind power prediction unit for predicting the wind power generation data for the next time period; and a photovoltaic power prediction unit for predicting the photovoltaic power generation data for the next time period. The prediction of power generation is a prior art, and those skilled in the art can accurately predict it.

[0079] The key point of this solution is how to avoid the battery charging and discharging simultaneously. Specifically: The battery function control module generates the charging time period and discharging time period of each battery based on the historical power generation data for the next time period, the time period for connecting to the grid, and the output power, and controls each battery to cycle charge and discharge. The battery function control module generates the charging time period and discharging time period of each battery based on the historical power generation data for the next time period, the time period for connecting to the grid, and the output power, and controls each battery to cycle charge and discharge.

[0080] For example, the historical power generation in the next 24 hours is calculated to be 100, and considering that the electricity price is the highest from 2pm to 4pm, it is expected to output 100 power during this time period. 100 power output requires 300 batteries to output at full power, and the remaining 200 batteries are idle. In this way, during the charging phase within 24 hours, the 300 batteries expected to be used for discharge are charged first, and the remaining 200 batteries are idle. When the discharge node is reached, the remaining 300 fully charged batteries are discharged, and the remaining 200 batteries are used to receive possible wind power or photovoltaic power.

[0081] Therefore, in this process, it is necessary to accurately predict the amount of electricity that each battery can release. In practice, although the amount of electricity stored in the battery can be roughly calculated based on the charging current, considering the loss during the charging process and the aging of the battery, the estimated amount of electricity needs to be corrected by a correction factor, which is actually the battery life score. For this reason, this solution uses the following method to calculate the life score:

[0082] The power prediction module includes: a data collection unit, a data preprocessing unit, a data prediction unit, and a charge and discharge performance calculation unit.

[0083] The data collection unit and data preprocessing unit are mainly used for data cleaning and preprocessing. The data prediction unit is used to calculate the expected lifespan, or the battery aging score. The charge and discharge performance calculation unit calculates how much power the battery can discharge or charge based on the expected lifespan of the battery.

[0084] refer to Figure 2 Specifically: the electricity consumption data includes: charge and discharge voltage and time curve, charge and discharge current and time curve, charge and discharge temperature and time curve.

[0085] The above data are the most critical data in battery life. Although the battery internal resistance is also important, there is a certain relationship between the battery internal resistance and the voltage and current. Its introduction will lead to low data redundancy. After many attempts, this solution directly eliminates it to reduce the overfitting of the model.

[0086] A data collection unit, used to collect power consumption data of each battery i, where i is the index of the battery;

[0087] The data preprocessing unit is used to preprocess the power consumption data of each battery i to obtain the prediction data H of each battery i ;

[0088] The data preprocessing unit includes: a data dimension reducer, a normalizer, and a correlation predictor.

[0089] Data dimensionality reducer for reducing the dimensionality of prediction data H i to reduce its dimensionality;

[0090] The preprocessing in this solution mainly reduces the dimensionality of the data to reduce its complexity. The reduced prediction data H i ={[t min(v1) , v1], [t min(v2) , v2], [t min(a1) , A1], [t min(a2) , A2], [t T1 , T1], [t T2 , T2]};

[0091] [t min(v1) , v1] = {SV1 s ≥V max}, s = 1, 2, 3…, t min(v1) is the time when the voltage of the battery first reaches the maximum value during charging, v1 is the voltage value when the battery terminal voltage is first not less than the maximum voltage value, s is the number of charging times, SV1 s is the voltage data of the charging voltage - time curve, and V max is the rated maximum charging voltage;

[0092] [t min(v2) , v2] = {SV2 d ≥V min}, d = 1, 2, 3…, t min(v2) is the time when the voltage of the battery first reaches the minimum value during discharging, v2 is the voltage value when the battery terminal voltage is first not greater than the minimum voltage value, d is the number of discharging times; SV2 s is the voltage data of the discharging voltage - time curve, and V min is the rated minimum charging voltage;

[0093] [t min(a1) , A1] = {SA2 s ≤A max}, s = 1, 2, 3…, t min(a1) is the time when the current of the battery starts to decline during charging, A1 is the current value when the battery current starts to decline, A max is the maximum current value during charging, s is the number of charging times; SA2 s is the current data of the charging current - time curve;

[0094] [t min(a2) , A2] = {SA2 d ≥Ax1}, d = 1, 2, 3…, t min(a2) is the time when the current of the battery starts to rise during discharging, A2 is the current value when the battery current starts to rise, Ax1 is the maximum current value during discharging, d is the number of discharging times; SA2d is the current data of the discharge current vs. time curve;

[0095] [t T1 , T1] = {ST1 s ≥ T1 max}, s = 1, 2, 3…, t T1 is the time when the temperature of the battery first reaches the maximum value during charging, T1 is the maximum temperature reached by the battery during charging, s is the number of charging times, ST1 s is the temperature data of the charging temperature vs. time curve, T1 max is the maximum charging temperature;

[0096] [t T2 , T2] = {ST1 d ≥ T2 max}, s = 1, 2, 3…, t T2 is the time when the temperature of the battery first reaches the maximum value during discharging, T2 is the maximum temperature reached by the battery during discharging, d is the number of discharging times, ST2 d is the temperature data of the discharging temperature vs. time curve, T2 max is the maximum discharging temperature.

[0097] The dimensionality reduction in this solution is actually to find the feature points where the information in the data of the battery has an offset. Compared with the redundant data information, these data information can better indicate the offset direction of the battery usage situation.

[0098] The normalizer is used to normalize the data; normalization is a prior art and will not be elaborated here.

[0099] The correlation predictor is used to screen the charge-discharge correlation of the data and screen out the data error prediction data

[0100] For the prediction data, calculate the correlation coefficient P between each data. If the range of the correlation coefficient between each data item is within the preset range, the prediction data is normal; if not, the battery is deleted.

[0101]

[0102] Among them, P X,Y represents the correlation coefficient between the data item and , and They respectively represent the X-th data and the Y-th data in the predicted data Hi. The value range of P is between -1 and 1, where 1 represents a complete positive correlation, -1 represents a complete negative correlation, and 0 represents no linear correlation. Therefore, in practice, only a reasonable threshold needs to be set to control the amount of data introduced. If the prediction requirement is high, the preset value can be set larger; otherwise, it can be set smaller. For the battery corresponding to the predicted data with inaccurate prediction, the maintenance flag is compared to let the workers process it regularly.

[0103] The above are the preprocessing method and screening method of the predicted data. After the preprocessing and screening are completed, for each battery, only its predicted input data needs to be input into the prediction unit.

[0104] Reference Figure 3 , the data prediction unit, and the predicted data of each battery are input into the CNN neural network model to obtain the predicted life of each battery;

[0105] Specifically, the data prediction unit includes: an input layer, a convolutional layer, a pooling layer, an Attention layer, and an output layer. The input layer is used to input the predicted data. The convolutional layer slides the convolutional kernel on the input predicted data H i to calculate the dot product of the convolutional kernel and the input data, generate a feature map, and generate the local features of the predicted data H i ; the pooling layer receives the output from the previous convolutional layer as input and reduces the feature dimension; the process of pooling is the process of reducing the dimension of the data. There are generally multiple convolutional layers and pooling layers. The implicit features obtained by multiple convolutional layers will ultimately be input into the fully connected layer for feature fusion. In this solution, the fully connected layer is replaced by the Attention layer.

[0106] Reference Figure 4 , the Attention layer performs weighted summation on the implicit features generated by each convolutional layer to generate the final predicted life score U.

[0107] In the Attention layer,

[0108] R ki = tanh(u k ·q ki ·u i );

[0109] Among them, R ki represents an intermediate variable, representing the attention weight between the k-th position and the i-th position. u k represents the k-th hidden state, q ki represents the weight coefficient matrix to be learned, and tanh represents the double tangent activation function;

[0110]

[0111] The pki represents the normalized attention weight, indicating the degree of attention of the k-th position to the i-th position;

[0112] R i represents the hidden variable at the k-th position.

[0113] Since the aging degrees of the batteries are not consistent, the magnitudes of the finally obtained predicted life scores U are not consistent. To improve the model accuracy, the loss function LS of the data prediction unit is as follows:

[0114]

[0115] where U i is the actual value of the predicted life score, is the predicted value of the predicted life score.

[0116] In this solution, the mean percentage error is adopted, which can accurately evaluate the prediction accuracy on different actual values and is independent of the dimension of the data. Therefore, it can effectively detect the errors of the model and improve the prediction accuracy.

[0117] The above is the basic structure of the model and the corresponding loss function, so that the training method and the training process of the model can be accurately represented. When optimizing the model, it is necessary to continuously update the weight parameters in the convolutional neural network. In the existing update methods, when dealing with the large-scale sample range in this solution, the update process of the weight parameters is slow, which greatly increases the training cost of the model. Therefore, in this solution, when updating the weight parameters, the following method is adopted:

[0118] Set the initial learning rate and step size α, the decay coefficients of the first-order moment Q 1 vector and the second-order moment Q 2 vector are set to L1 and L2, and the loss function LS is used to optimize the weight parameter θ. The optimization process is as follows:

[0119]

[0120] where Q 1 is the first-order moment vector, used to smooth the gradient and reduce the influence of noise on the gradient direction. Q 2 is the second-order moment vector, used to adaptively adjust the learning rate of each parameter. t is the time step, representing the current iteration number. b is a non-zero constant. E t is the gradient of the objective function at the current time step with respect to the parameter θ.

[0121] The charge-discharge performance calculation unit converts the predicted lifespan into a corresponding conversion ratio, calculates the current standard charge-discharge amount based on the charge-discharge data, and multiplies the conversion ratio by the standard charge-discharge amount to calculate the charge-discharge capacity of each battery.

[0122] How the charge-discharge performance calculation unit generates the conversion ratio only requires statistical analysis, and the specific method will not be elaborated here.

[0123] The above description is only some preferred embodiments of the present application and an explanation of the applied technical principles. Those skilled in the art should understand that the scope of the invention involved in the embodiments of the present application is not limited to the technical solutions formed by the specific combination of the above technical features, and should also cover other technical solutions formed by any combination of the above technical features or their equivalent features without departing from the above inventive concept. For example, the technical solutions formed by mutually replacing the above features with the (but not limited to) technical features with similar functions disclosed in the embodiments of the present application.

Claims

1. A smart management system for an energy storage power station, comprising: A photovoltaic monitoring module for obtaining photovoltaic generation data of a photovoltaic power generation module; A wind energy monitoring module for obtaining wind power generation data of a wind power generation module; A energy storage monitoring module for detecting the power consumption data of each battery in the energy storage module; A grid connection control module for pre-inputting the time period when the microgrid is connected to the grid and the output power; An operation monitoring module for intelligent alarm of operation status, intelligent charge and discharge management, loss analysis, and power statistics analysis; It is characterized in that: it further includes: A power generation prediction module for calculating the power generation data of the next time period according to the photovoltaic generation data and the wind power generation data; A power prediction module for generating the power of each battery based on the power consumption data to obtain the charge and discharge capacity of each battery; A battery function control module for generating the charging time period and the discharging time period of each battery based on the power generation data of the next time period, the time period of connecting to the grid, and the output power, and controlling the cyclic charge and discharge of each battery; The power prediction module includes: A data collection unit for collecting the power consumption data of each battery i, where i is the index of the battery; A data preprocessing unit for preprocessing the power consumption data of each battery i to obtain the prediction data H of each battery i ; A data prediction unit for inputting the prediction data of each battery into a CNN neural network model to obtain the predicted life of each battery; A charge and discharge performance calculation unit for converting the predicted life into a corresponding conversion ratio, calculating the current standard charge and discharge amount according to the charge and discharge data, and multiplying the conversion ratio by the standard charge and discharge amount to calculate the charge and discharge capacity of each battery.

2. The intelligent management system for an energy storage power station according to claim 1, characterized in that: A battery function control module for calculating the number of batteries that need to be charged in the future according to the power generation data of the next time period, and calculating the maximum output power in the future time period according to the number of batteries that need to be charged in the future.

3. The intelligent management system for an energy storage power station according to claim 1, characterized in that: The power generation prediction module includes: A wind power prediction unit for predicting the wind power generation data of the next time period; A photovoltaic prediction unit for predicting the photovoltaic generation data of the next time period.

4. The intelligent management system for an energy storage power station according to claim 1, characterized in that: The power consumption data includes: charge and discharge voltage vs. time curve, charge and discharge current vs. time curve, charge and discharge temperature vs. time curve.

5. The intelligent management system for an energy storage power station according to claim 1, wherein: The data preprocessing unit includes: Data dimension reducer for reducing the dimension of prediction data H i for dimension reduction; A normalizer for normalizing the data; A correlation predictor for screening the charge and discharge correlation of the data and screening out the data error prediction data.

6. The intelligent management system for an energy storage power station according to claim 5, wherein: The predicted data H after dimensionality reduction i ={[t min(v1) , v1], [t min(v2) , v2], [t min(a1) , A1], [t min(a2) , A2], [t T1 , T1], [t T2 , T2]}; [t min(v1) , v1] = {SV1 s ≥V max}, s = 1, 2, 3…, t min(v1) is the time when the voltage first reaches the maximum value during battery charging, v1 is the voltage value when the battery terminal voltage is first not less than the maximum voltage value, s is the number of charging times, SV1 s is the voltage data of the charging voltage vs. time curve, V max is the rated maximum charging voltage; [t min(v2) , v2] = {SV1 d ≥V min}, d = 1, 2, 3…, t min(v2) is the time when the voltage of the battery first reaches the minimum value during discharge, v2 is the voltage value when the battery terminal voltage is first not greater than the minimum voltage value, and d is the number of discharge times; SV1 d is the voltage data of the discharge voltage vs. time curve, and V min is the minimum value of the rated charging voltage; [t min(a1) ,A1] = {SA2 s ≤ A max}, s = 1, 2, 3…, t min(a1) is the time when the current starts to decrease during battery charging, A1 is the current value when the battery current starts to decrease, A max is the maximum current value during charging, s is the number of charging times; SA2 s is the current data of the charging current vs. time curve; [t min(a2) , A2] = {SA2 d ≥ Ax1}, d = 1, 2, 3…, t min(a2) is the time when the current starts to rise during battery discharge, A2 is the current value when the battery current starts to rise, Ax1 is the maximum current value during discharge, d is the number of discharge cycles; SA2 d is the current data of the discharge current vs. time curve; [t T1 , T1] = {ST1 s ≥ T1 max}, s = 1, 2, 3…, t T1 is the time when the temperature first reaches the maximum value during battery charging, T1 is the highest temperature reached during battery charging, s is the number of charging times, ST1 s is the temperature data of the charging temperature data and time curve, T1 max is the maximum charging temperature; [t T2 , T2] = {ST1 d ≥ T2 max}, s = 1, 2, 3…, t T2 is the time when the temperature of the battery first reaches the maximum value during discharge, T2 is the highest temperature reached during battery discharge, d is the number of discharges, ST1 d is the temperature data of the discharge temperature data and time curve, T2 max is the maximum value of the discharge temperature.

7. The intelligent management system for an energy storage power station according to claim 6, characterized in that: For the prediction data, calculate the correlation coefficient P between each data. If the range of the correlation coefficient between each data item is within the preset range, the prediction data is normal. If it is not within the preset range, delete the battery; where, P X,Y represents the correlation coefficient between and and respectively represent the Xth data item and the Yth data item in the predicted data Hi.

8. The intelligent management system for an energy storage power station according to claim 6, wherein: The data prediction unit includes: Convolutional layer, sliding a convolutional kernel over the input prediction data H i to calculate the dot product of the convolutional kernel and the input data, generating a feature map to generate local features of the prediction data H i ; A pooling layer that receives the output from the previous convolutional layer as input and reduces the feature dimension; An Attention layer that performs weighted summation on the hidden features generated by each convolutional layer to generate the final predicted life score U.

9. The intelligent management system for an energy storage power station according to claim 8, characterized in that: In the Attention layer, R ki =tanh(u k ·q ki ·u i ); Among them, R ki represents an intermediate variable, representing the attention weight between the k-th position and the i-th position, u k represents the k hidden states, q ki represents the weight coefficient matrix to be learned, and tanh represents the hyperbolic tangent activation function; pki represents the normalized attention weight, indicating the degree of attention of the k-th position to the i-th position; R i represents the hidden variable at the k-th position.

Images