Distributed finite control set model prediction control method and controller

Abstract

The invention discloses a distributed finite control set model prediction control method and a controller, and the method comprises the steps: converting the voltage of a first bridge arm into a voltage series form for expression, and constructing a first bridge arm discrete mathematical model expressed by the voltage series; calculation is carried out based on a first bridge arm voltage discrete mathematical model expressed by voltage levels, selection of switching states is converted into selection of voltage levels, the number of switching states needing to be calculated is reduced, the calculation burden is reduced, and therefore rapid control over the modular multilevel matrix converter is achieved. Moreover, potential categories of false data injection attacks possibly occurring in the control structure are further analyzed, the defects of a conventional detector are explored, and a detector combining a fuzzy logic structure and a neural network is provided to achieve rapid detection of the false data injection attacks and ensure system safety.

Description

Technical Field

[0001] The present application relates to the technical field of power systems, and in particular to a distributed finite control set model predictive control method and controller. Background Art

[0002] Since the MMMC (Modular Multilevel Matrix Converter) has a total of 9 bridge arms, each bridge arm is composed of a number of AC-AC conversion sub-modules connected in series. In the related technology, the traditional MPC (model predictive control) control system based on the switch state is complex to calculate, and it needs to traverse the switch state of all sub-modules in each bridge arm during operation, resulting in a huge calculation burden for the model predictive control of the MMMC. Summary of the invention

[0003] In order to solve the deficiencies of the prior art, this application adopts the following technical solutions: In a first aspect, the present application provides a distributed finite control set model predictive control method, which is applied to the control of a modular multi-level matrix converter, wherein the modular multi-level matrix converter includes nine bridge arms, each bridge arm includes a plurality of AC-AC conversion sub-modules, and each bridge arm is arranged between an input port and an output port of the modular multi-level matrix converter, and the predictive control method includes: Based on Kirchhoff's current law, the dynamics of the first bridge arm current are modeled to obtain a bridge arm current dynamic model of the first bridge arm; The voltage of the first bridge arm is expressed in the form of a voltage series, and based on the first bridge arm voltage expressed in the voltage series, the bridge arm current dynamic model is discretized to construct a discrete mathematical model of the first bridge arm, wherein the discrete mathematical model includes the first bridge arm voltage expressed in the voltage series; Based on the discrete mathematical model, according to the limited voltage level of each AC-AC conversion submodule in the first bridge arm, respectively calculate the predicted output value of the first bridge arm at different voltage levels; Obtaining a reference output value of the first bridge arm, and constructing a basic cost function corresponding to the first bridge arm based on the predicted output value of the first bridge arm and the reference output value, and taking the minimum of the basic cost function as the optimal switching state; All bridge arms are traversed, and the optimal switch state of each bridge arm is selected to act on the modular multi-level matrix converter.

[0004] In summary, the present application provides a distributed finite control set module predictive control method, which converts the voltage of the first bridge arm into a voltage series form, constructs a discrete mathematical model of the first bridge arm expressed in voltage series, and performs calculations based on the discrete mathematical model of the first bridge arm voltage expressed in voltage series. By converting the selection of switching states into the selection of voltage levels, the number of switching states that need to be calculated is reduced, and the calculation burden is reduced, thereby achieving rapid control of the modular multi-level matrix converter.

[0005] Further, the voltage level of the first bridge arm is expressed as follows: ; ; ; Where V bk Represents the voltage of the kth bridge arm, N b Represents the discretized voltage level, N level represents the number of non-negative voltage levels that can be selected after considering the balance between computation time and accuracy, V k-s Indicates the voltage amplitude corresponding to each voltage level, V c-kj Represents the capacitor voltage of the jth AC-AC conversion submodule in the first bridge arm.

[0006] Furthermore, the discrete mathematical model is expressed by the following formula: ; Where, T s is the sampling time, Indicates the bridge arm viewed from the outside Voltage drop, L b represents the bridge arm inductance, i bk (k) represents the current of the bridge arm at time k, V com (k) represents the potential difference between the neutral points on both sides at time k.

[0007] Furthermore, the method further comprises: The capacitor voltage of each AC-AC conversion submodule in the first bridge arm is obtained, and the capacitor voltage balance of the i-th AC-AC conversion submodule in the first bridge arm is achieved by the following equation: ; In the formula, is the control signal input to the modulation stage, is the gain parameter to be selected, sign(*) indicates the sign function, ibk indicates the current of bridge arm k, N indicates the number of AC-AC conversion submodules in each bridge arm, V c-kjrepresents the capacitor voltage of the jth AC-AC conversion submodule in the first bridge arm, V con is the bridge arm voltage reference.

[0008] Furthermore, the basic cost function is the square of the error between the predicted output value of the first bridge arm and the reference output value.

[0009] Furthermore, the method further comprises: Using a Kalman filter to detect any of the AC-AC conversion submodules to obtain a priori estimates and measurements of the AC-AC conversion submodule; Constructing a single-layer neural network, including configuring network parameters, wherein the network parameters at least include the number of nodes in the input layer, the hidden layer, and the output layer; Using the a priori estimate, the integral of the a priori estimate, the measured value and the integral of the measured value as inputs of the single-layer neural network, and judging whether the modular multi-level matrix converter is abnormal based on the output of the single-layer neural network; The output of the hidden layer is expressed by the following formula: ; The output of the output layer is expressed by the following formula: ; In the formula, represents the activation function, represents the i-th input node of the system, represents the input bias, J and K represent the number of input nodes and hidden nodes respectively, represents the output of the jth hidden layer neuron.

[0010] Furthermore, the method further comprises: Using a random weight method, obtaining a preset number of weights from an input layer to a hidden layer in the single-layer neural network, and a bias corresponding to the weight; The validity function is used to evaluate the weights obtained each time, and the optimal weights and corresponding biases are selected therefrom.

[0011] Furthermore, the validity function is expressed by the following formula: ; ; in, represents the validity function, which is used to evaluate the random input weights, is the intermediate variable, represents the error of the kth neuron, and M are learning parameters, and , the superscript T indicates transpose.

[0012] Furthermore, the method further comprises: Through forward propagation calculation, the weights of the output layer are optimized by minimizing the following objective function, which is expressed by the following formula: ; In the formula, represents the output layer weight, represents the weight corresponding to the jth neuron of the i-th sampling, Represents the output of the i-th sampling.

[0013] In a second aspect, the present application also provides a controller for a modular multi-level matrix converter, which adopts the distributed finite control set model predictive control method described in any one of the above items. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] Figure 1 A schematic diagram of a topological structure of a modular multi-level matrix converter provided in one embodiment of the present application; Figure 2 A flowchart of the steps of quickly controlling a modular multi-level matrix converter using a distributed finite control set model predictive control method provided by an embodiment of the present application; Figure 3 A schematic diagram of simulation results of a classic MPC method provided as a comparative example of the present application; Figure 4 A schematic diagram of simulation results of a sequential MPC method provided as a comparative example of the present application; Figure 5 A schematic diagram of simulation results of a distributed finite control set model predictive control method provided by an embodiment of the present application; Figure 6 A schematic diagram of a strategy for applying a distributed finite control set model predictive control method provided in one embodiment of the present application to a bridge arm; Figure 7 A flowchart of the steps of detecting Xu Jian data injection attack by a distributed finite control set model predictive control method provided by an embodiment of the present application; Figure 8 A schematic diagram of constructing a single-layer neural network in a distributed finite control set model predictive control method provided in one embodiment of the present application; Fig. 9 A diagram showing the detection results of a distributed finite control set model predictive control method provided by an embodiment of the present application under two false data injection attacks; Fig.10 A schematic diagram of the overall detection framework of a distributed finite control set model predictive control method provided in one embodiment of the present application. DETAILED DESCRIPTION

[0015] The present application will be described in detail below in conjunction with the specific implementation modes shown in the accompanying drawings, but these implementation modes do not limit the present application. Structural, methodological, or functional changes made by ordinary technicians in the field based on these implementation modes are included in the protection scope of the present application.

[0016] In order to solve the deficiencies of the prior art, the present application provides a distributed finite control set model predictive control method, which is applied to the control of a modular multi-level matrix converter. The modular multi-level matrix converter includes nine bridge arms, each of which includes a plurality of AC-AC conversion sub-modules, and each bridge arm is arranged between an input port and an output port of the modular multi-level matrix converter. The topological structure of the MMMC is as follows Figure 1 As shown, the MMMC consists of 9 bridge arms, each of which is connected between an input port and an output port.

[0017] like Figure 2 As shown, the predictive control method includes: Step S11, based on Kirchhoff's current law, modeling the dynamics of the first bridge arm current to obtain a bridge arm current dynamic model of the first bridge arm; Step S12, expressing the voltage of the first bridge arm in the form of a voltage series, based on the first bridge arm voltage expressed in the voltage series, discretizing the bridge arm current dynamic model, and constructing a discrete mathematical model of the first bridge arm, wherein the discrete mathematical model includes the first bridge arm voltage expressed in the voltage series; Step S13, based on a discrete mathematical model, according to the limited voltage levels of each AC-AC conversion submodule in the first bridge arm, respectively calculating the predicted output value of the first bridge arm at different voltage levels; Step S14, obtaining a reference output value of the first bridge arm, constructing a basic cost function corresponding to the first bridge arm based on the predicted output value and the reference output value of the first bridge arm, and taking the minimum basic cost function as the optimal switching state; Step S15, traversing all bridge arms, selecting the optimal switch state of each bridge arm to act on the modular multi-level matrix converter.

[0018] like Figure 3 , Figure 4 and Figure 5 As shown in the figure, the classical MPC method, the sequential MPC method and the distributed MPC method proposed in this paper are compared, and the error variable is introduced to analyze the results. Figure 3 , Figure 4 and Figure 5In the figure, part (a) shows the output three-phase current waveform, part (b) shows the source side three-phase current waveform, part (c) shows the submodule capacitor voltage, and part (d) shows the fast Fourier transform analysis result. εi Indicates the average tracking error percentage of the three-phase current, εc It represents the error percentage between the average capacitor voltage of a single bridge arm and the reference capacitor voltage. In the figure, the first bridge arm is selected for testing. THD is introduced to evaluate the current.

[0019] Based on the comparison results, the control structure proposed in this paper shows better performance in both current tracking and THD indicators by introducing more equivalent levels in the calculation. In addition, the control structure proposed in this paper has obvious advantages in the capacitor voltage balance control of the AC-AC conversion submodule. The calculation time of the classic MPC method is 20.3 seconds, the calculation time of the sequence MPC method is 6.4 seconds, and the calculation time of the distributed MPC method is 3.1 seconds. The calculation time of the distributed control scheme is significantly less than the two methods, and has a low average voltage error of 0.28%.

[0020] According to the above description, the present application provides a distributed finite control set module predictive control method, which converts the voltage of the first bridge arm into a voltage series form, constructs a discrete mathematical model of the first bridge arm expressed in voltage series, and performs calculations based on the discrete mathematical model of the first bridge arm voltage expressed in voltage series. By converting the selection of switching states into the selection of voltage levels, the number of switching states that need to be calculated is reduced, and the calculation burden is reduced, thereby achieving rapid control of the modular multi-level matrix converter.

[0021] like Figure 1 As shown, , and represents the voltage input to the grid side, and Indicates the three-phase current input to the grid side. Indicates the three-phase voltage on the output side, Represents the output three-phase voltage. The potential difference between the neutral points on both sides is defined as .

[0022] from Figure 1 It can be seen that each bridge arm is composed of N AC-AC conversion submodules. Assume that the capacitance of each AC-AC conversion submodule is , the reference voltage of each AC-AC conversion submodule is , then under ideal conditions, each AC-AC conversion submodule can output three voltage levels: , 0 and The inductance of the bridge arm is denoted by .

[0023] In step S11 of the embodiment of the present application, the dynamics of the first bridge arm current are modeled by applying Kirchhoff's current law to the first bridge arm current, thereby obtaining a bridge arm current dynamic model of the first bridge arm. The modeling process can be expressed by the following formula: (1); Where V u 、V v and V w Indicates the voltage input to the grid side, V r 、V s and V t Represents the output three-phase voltage, V com Represents the potential difference between the neutral points on both sides, I 3x3 represents the identity matrix of dimension 3, L b represents the inductance of the bridge arm, i bk Indicates the b k Bridge arm current, k={1, 2, …, 9}, V bk Indicates the b k No. bridge arm voltage.

[0024] Combining formula (1) and Figure 1 In the structure, the currents on the input and output sides satisfy the following relationship: (2); In the formula, Indicates the three-phase current input to the grid side. Indicates the three-phase voltage on the output side.

[0025] In the related art, in the traditional MPC control algorithm based on the switch state, the switch state of each AC-AC conversion submodule of the first bridge arm is represented as ,in , then the voltage of any bridge arm is expressed by the following formula: (3); In the formula, represents the sum of the voltages of all AC-AC conversion submodules in the kth bridge arm, V kj Indicates the switch status of the AC-AC conversion submodule, Represents the capacitor voltage of the jth AC-AC conversion submodule in the kth bridge arm.

[0026] Substituting formula (3) into formula (1) and discretizing it, the dynamics of bridge arm k at time k can be expressed as: (4); Where, L b represents the inductance of the bridge arm, T s represents the sampling time, Represents the voltage drop of the kth bridge arm from an external perspective.

[0027] At time k, the bridge arm current at time (k+1) can be predicted using equation (4). Assuming that the reference current of each bridge arm has been obtained, the basic cost function of each bridge arm can be constructed.

[0028] In the traditional switch state-based MPC control algorithm, the main idea is to traverse all switch states to optimize the basic cost function of each bridge arm, and then select the switch state that minimizes the current error at time k+1. The total number of switch states that need to be calculated during each traversal is It can be seen that the number of iterations in the traditional switch-state-based MPC control algorithm increases exponentially with the number of sub-modules, and the amount of calculation is huge.

[0029] In step S12 of the present application, the voltage of the first bridge arm is expressed in the form of a voltage level. As an implementation method, the voltage level of the first bridge arm is expressed in the following form: (5); In the formula, represents the sum of the voltages of all AC-AC conversion submodules in the kth bridge arm, N b Represents the discretized voltage level, V k-s Indicates the voltage amplitude corresponding to each voltage level.

[0030] Furthermore, N b The expression is as follows: (6); Where N b represents the discretized voltage level, Represents the number of non-negative voltage levels that can be selected after considering the balance between computation time and accuracy.

[0031] When the capacitor voltage of the AC-AC conversion submodule does not change significantly within a single time step, the measurement information of the previous time step can be used to calculate V k-s , the calculation formula is expressed as follows: (7); In the formula, Indicates the voltage amplitude corresponding to each level, V c-kj represents the capacitor voltage of the jth AC-AC conversion submodule in the first bridge arm, Represents the number of non-negative voltage levels that can be selected after considering the balance between computation time and accuracy.

[0032] Based on the voltage of the first bridge arm expressed by the voltage series, the bridge arm current dynamic model is discretized to construct a discrete mathematical model of the first bridge arm. The discrete mathematical model includes the voltage of the first bridge arm expressed by the voltage data. The discrete mathematical model is expressed by the following formula: (8); Where, T s represents the sampling time, L b represents the inductance of the bridge arm, Indicates the bridge arm viewed from the outside The voltage drop, V com Represents the potential difference between the neutral points on both sides.

[0033] In the above manner, the selection of the switch state of the AC-AC conversion submodule can be converted into the selection of the voltage level, thereby reducing the number of switch states that need to be calculated. Reduce to .

[0034] As an implementation method, in step S14, based on the predicted output value and the reference output value of the first bridge arm, a basic cost function corresponding to the first bridge arm is constructed. The basic cost function is the square of the error between the predicted output value and the reference output value of the first bridge arm, and the formula is expressed as follows: (9); In the formula, Indicates the first bridge arm exist The reference output value at the moment, Indicates the first bridge arm exist The predicted output value at time .

[0035] Based on the finite voltage level of the bridge arm, the basic cost function of each bridge arm in formula (9) is optimized, and the switching state that minimizes the current error of each bridge arm is selected as the optimal switching state. The optimal switching state is applied to the modular multilevel matrix converter, thereby realizing rapid control of the modular multilevel matrix converter.

[0036] Furthermore, the effectiveness of the control method provided by the present application depends on the modulation stage and the capacitor voltage balance of the AC-AC conversion submodule. As an implementation method, the control method provided by the present application also includes obtaining the capacitor voltage of each AC-AC conversion submodule in the first bridge arm, and realizing the capacitor voltage balance of the i-th AC-AC conversion submodule in the first bridge arm through the following equation: (10); In the formula, represents the control signal input to the modulation stage, Indicates the gain parameter to be selected, sign(*) indicates the sign function, V c-ki represents the capacitor voltage of the i-th AC-AC conversion submodule in the first bridge arm, V c-kj represents the capacitor voltage of the jth AC-AC conversion submodule in the first bridge arm, V con Indicates the bridge arm voltage reference.

[0037] The strategy of a single bridge arm is as follows Figure 6 As shown, and the same configuration is applied to all other bridge arms, the central controller is responsible for handling higher-level tasks such as overall energy balance and current control, while the distributed controller achieves more efficient management of the energy balance of the AC-AC conversion submodule, and achieves the balance of the capacitor voltage of the modulation stage and the AC-AC conversion submodule. For the problem of submodule capacitor voltage balance solved by this application, it is assumed that all LCs (local controllers) in the same bridge arm can communicate effectively, so they will be able to obtain the capacitor voltage information of other SMs (submodules) in the bridge arm. Under the above assumptions, it is assumed that the bridge arm voltage reference transmitted by the high-level controller is Since FCS-MPC omits the modulation stage, the capacitor voltage balance of the i-th SM in the k-th bridge arm can be achieved by equation (10). According to equation (10), the control signal for achieving capacitor voltage balance can be obtained by separate calculation in each AC-AC conversion submodule, and the parallel structure of multiple subcontrollers can reduce the calculation time.

[0038] Furthermore, in the distributed finite control set model predictive control method provided by this application, considering that the submodules under distributed control rely on the transmission network for data transmission, there is a potential risk of false data injection attack, and in some cases, the existing Kalman filter cannot detect the attack. In this regard, this application further analyzes the detectability of FDIA (false data injection attack) in the MMMC system and provides corresponding detection means to ensure that the system has the detection capability for a variety of different FDIA attacks to ensure the stability of the MMMC system.

[0039] Specifically, this application analyzes the detectability of FDIA in the MMMC system as follows: False data injection attacks may cause the local controller to receive incorrect information. Since each AC-AC conversion submodule is decoupled, the capacitor voltage state matrix of the cluster is a diagonal matrix. For a single AC-AC conversion submodule, its state equation can be written in scalar form as follows: (11); In the formula, represents the capacitor voltage, represents the control input, w(k) represents the process noise, v(k) represents the measurement noise, and w(k) satisfies , v(k) satisfies .

[0040] When a false data injection attack occurs, the measured values ​​of the capacitor voltage will be changed first, which will cause the control results based on these measured values ​​to deviate. is the attack sequence. Under the attack, the state equation of the system can be expressed as: (12); In the formula, represents the attack sequence, Indicates the system control input after being attacked, Indicates the system output after being attacked. Indicates the system status after being attacked.

[0041] The Kalman filter detects the AC-AC conversion submodule to obtain the prior estimation and measurement value of the AC-AC conversion submodule. The detection process can be expressed as follows: (13); In the formula, Represents the predicted state, represents the prediction covariance matrix, represents the Kalman gain, Indicates the corrected system status, represents the corrected covariance, Indicates the measurement output.

[0042] definition is the error between the estimated value and the reference value, is the state estimation error, is the residual, which can be expressed as: (14); In the formula, Indicates the actual voltage value of the submodule.

[0043] Combining formula (13) and formula (14), we can get: is the same as e(k), and the state matrix of the AC-AC conversion submodule does not contain unstable eigenvalues. While z(k) remains finite. Since there is no capacitor voltage reference value in the local controller, the false data injection attack on the MMMC system can be simplified to the following form: (15); In the formula, represents the error between the estimated value and the reference value, Represents the residual.

[0044] The existing Kalman filter detection method mainly relies on residual-based monitoring, which is expressed as follows: (16); In the formula, Represents the detection function.

[0045] At each sampling moment, Gi is calculated. When Gi exceeds the preset threshold When , the detector triggers an alarm, indicating that a false data injection attack is detected. At this time, the preset threshold is set to ,in , and design , so that the detector does not trigger an alarm at any time. Before the false data injection attack occurs, the parameters P(k) and K(k) of the Kalman filter have converged, that is: (17); Where P(k) represents the covariance matrix and K(k) represents the Kalman gain.

[0046] Based on the above description, the system equation of the residual under the false data injection attack can be derived as follows: (18); In the formula, represents the residual, Indicates the attack sequence.

[0047] Combining formula (15) and formula (16), we can draw the following conclusion: If no alarm is triggered at any time k, then satisfy Assumptions , we can observe . Thus, the following conclusions can be derived: (19); In the formula, Indicates the detection threshold.

[0048] Set the start and end time of the false data injection attack to , and use Represents the attack sequence. Based on constraint (19), a simple and Kalman filter-undetectable covert attack algorithm can be designed, whose characteristic is that the gradient satisfies the requirements of formula (19).

[0049] Based on the above analysis of the attack forms, Figure 7As shown, the distributed finite control set model predictive control method provided by the present application also includes the following steps to efficiently detect a variety of different FDIA attacks including the false data injection attack described above, including the following steps: Step S21, using a Kalman filter to detect any AC-AC conversion submodule to obtain a priori estimation and measurement value of the AC-AC conversion submodule; Step S22, constructing a single-layer neural network, including configuring network parameters, the network parameters at least including the number of nodes in the input layer, the hidden layer, and the output layer; Step S23, taking the a priori estimate, the integral of the a priori estimate, the measured value and the integral of the measured value as inputs of a single-layer neural network, and judging whether the modular multi-level matrix converter is abnormal based on the output of the single-layer neural network.

[0050] The distributed finite control set model predictive control method provided in this application directly obtains the detection result by processing the output of the Kalman filter. Figure 8 As shown in the figure, a single-layer neural network is constructed and the network parameters are configured. is the activation function, h is the output of the hidden layer, represents the weight matrix, and b is the bias of the input layer. The network parameters include at least the number of nodes in the input layer, hidden layer, and output layer. The output of the hidden layer is expressed by the following formula: (20); In the formula, represents the jth hidden layer, represents the activation function, represents the i-th input node of the system, Represents the input bias, and the output of the output layer is expressed by the following formula: (twenty one); In the formula, represents the output of the output layer, J and K represent the number of input nodes and hidden nodes respectively, represents the output of the jth hidden layer neuron.

[0051] The AC-AC conversion submodule is detected by the Kalman filter to obtain the prior estimation and measurement value of the AC-AC conversion submodule. The prior estimation, the integral of the prior estimation, the measurement value and the integral of the measurement value are used as the input of the single-layer neural network. Through the calculation of the single-layer neural network, it is judged whether the modular multi-level matrix converter is abnormal based on the output result of the single-layer neural network. The input sequence length for attack detection is set to N, denoted as x, and The corresponding test results are ,in ,when , it indicates that a false data injection attack has occurred. , it indicates that no false data injection attack has occurred.

[0052] like Fig. 9 As shown in Figure 2, the proposed detector is simulated and verified under the step false data injection attack and the covert false data injection attack proposed in this paper. The AC-AC conversion submodule capacitor voltage value received by the local controller and the AC-AC conversion submodule capacitor voltage true value when the two attacks occur are shown in Figure 2. Fig. 9 (a) and Fig. 9 As shown in (b), it can be seen that the false data injection attack can cause a large gap between the information obtained by the local controller and the true value, resulting in a decrease in the system control effect. The results of the two detectors are shown in Fig. 9 As shown in (c), the conventional Kalman detector is slightly faster than the proposed detector in detecting step false data injection attacks due to its simple structure, but it cannot detect the covert attacks in this article. The detector proposed in this application can effectively detect both attacks.

[0053] As an implementation method, step S23 also includes: Step S231, using a random weight method to obtain a preset number of weights from the input layer to the hidden layer in a single-layer neural network, and a bias corresponding to the weight; Step S232, using the validity function to evaluate the weights obtained each time, and selecting the optimal weight and corresponding bias.

[0054] Specifically, in step S231, the weight of the input layer is obtained using a random weight method to reduce the computational burden of online detection. The random weight method is expressed as follows: (twenty two); In the formula, represents the weight, and Represents the scale parameters for weights and biases.

[0055] If the number of neurons in the output layer is set to L, the output error is defined as follows: (twenty three); In the formula, Represents the sampled value output within N steps.

[0056] In step S232, for each hidden layer neuron, the input weight is randomly generated. Second, by selecting the validity function to evaluate, thus ensuring that the randomly selected input layer weights can converge quickly for training.

[0057] As an implementation method, the selected validity function is expressed as follows: (twenty four); In the formula, represents the validity function, which is used to evaluate the random input weights, is the intermediate variable, represents the error of the kth neuron, and M are learning parameters, and , the superscript T indicates transpose.

[0058] in: (25); In the formula, represents transpose, and M is a learning parameter.

[0059] As an implementation method, the weights of the output layer are optimized by minimizing the objective function through forward propagation calculation.

[0060] Specifically, through Select the best of the randomly generated values , its optimized form is: (26); In the formula, represents the optimal input bias.

[0061] After obtaining the input layer weights, the output layer weights Through forward propagation calculation, the optimization is performed by minimizing the following objective function: (27); In the formula, represents the output layer weight, represents the weight corresponding to the jth neuron of the i-th sampling, Represents the output of the i-th sampling.

[0062] By solving this optimization problem: (28); By selecting a suitable sample data set and training according to the above method, real-time detection of the AC-AC conversion table module can be achieved. In the above description, the input of the single-layer neural network input layer is selected as ,in represents the state estimated by the Kalman filter (prior estimate), and the output of the single-layer neural network is .like Fig.10As shown, the accuracy of the online detection capability of the control method can be further improved by adjusting parameters and training under a variety of different FDIA attacks.

[0063] According to the above description, the present application provides a distributed finite control set module predictive control method, which converts the voltage of the first bridge arm into a voltage series form, constructs a discrete mathematical model of the first bridge arm expressed in voltage series, and performs calculations based on the discrete mathematical model of the first bridge arm voltage expressed in voltage series. By converting the selection of the switch state into the selection of the voltage level, the number of switch states that need to be calculated is reduced, and the calculation burden is reduced, thereby realizing rapid control of the modular multi-level matrix converter; and, by analyzing the false data injection attack target and constraints of the Kalman filter, a single-layer neural network is constructed to improve the comprehensiveness of the false data injection attack detection and ensure the stable operation of the modular multi-level matrix converter.

[0064] In a second aspect, the present application also provides a controller for a modular multilevel matrix converter, which adopts the distributed finite control set model predictive control method described above to achieve rapid control of the modular multilevel matrix converter and ensure stable operation of the modular multilevel matrix converter.

[0065] It will be appreciated that the word "exemplary" as used herein means "serving as an example, instance, or illustration". Any embodiment described as "exemplary" is not necessarily preferred or superior to other embodiments and / or does not exclude the combination of features of other embodiments. It will be appreciated that certain features of the present application described in the context of separate embodiments for the sake of clarity may also be provided in a single embodiment by combination. Conversely, various features of the present application described in the context of a single embodiment for the sake of clarity may also be provided individually or in any suitable combination or as any other described embodiment of the present application.

[0066] The above disclosure is only the preferred embodiment of the present application, but it is not intended to limit the scope of rights of the present application. A person of ordinary skill in the art can understand that without departing from the spirit and scope of the present application and the appended claims, changes, modifications, substitutions, combinations, and simplifications should all be equivalent replacement methods and still fall within the scope of the invention.

Claims

1. A distributed finite control set model predictive control method, applied to the control of a modular multi-level matrix converter, wherein the modular multi-level matrix converter comprises nine bridge arms, each bridge arm comprises a plurality of AC-AC conversion submodules, and each bridge arm is arranged between an input port and an output port of the modular multi-level matrix converter, characterized in that: The predictive control method comprises: Based on Kirchhoff's current law, the dynamics of the first bridge arm current are modeled to obtain a bridge arm current dynamic model of the first bridge arm; The voltage of the first bridge arm is expressed in the form of a voltage series, and based on the first bridge arm voltage expressed in the voltage series, the bridge arm current dynamic model is discretized to construct a discrete mathematical model of the first bridge arm, wherein the discrete mathematical model includes the first bridge arm voltage expressed in the voltage series; Based on the discrete mathematical model, according to the limited voltage level of each AC-AC conversion submodule in the first bridge arm, respectively calculate the predicted output value of the first bridge arm at different voltage levels; Obtaining a reference output value of the first bridge arm, and constructing a basic cost function corresponding to the first bridge arm based on the predicted output value of the first bridge arm and the reference output value, and taking the minimum of the basic cost function as the optimal switching state; All bridge arms are traversed, and the optimal switch state of each bridge arm is selected to act on the modular multi-level matrix converter.

2. The distributed finite control set model predictive control method according to claim 1, characterized in that: The voltage level of the first bridge arm is expressed as follows: ； ； ； Where V bk Represents the voltage of the kth bridge arm, N b Represents the discretized voltage level, N level represents the number of non-negative voltage levels that can be selected after considering the balance between computation time and accuracy, V k-s Indicates the voltage amplitude corresponding to each voltage level, V c-kj Represents the capacitor voltage of the jth AC-AC conversion submodule in the first bridge arm.

3. The distributed finite control set model predictive control method according to claim 2, characterized in that: The discrete mathematical model is expressed by the following formula: ； Where, T s is the sampling time, Indicates the bridge arm viewed from the outside Voltage drop, L b represents the bridge arm inductance, i bk (k) represents the current of the bridge arm at time k, V com (k) represents the potential difference between the neutral points on both sides at time k.

4. The distributed finite control set model predictive control method according to claim 3, characterized in that: The method further comprises: The capacitor voltage of each AC-AC conversion submodule in the first bridge arm is obtained, and the capacitor voltage balance of the i-th AC-AC conversion submodule in the first bridge arm is achieved by the following equation: ； In the formula, is the control signal input to the modulation stage, is the gain parameter to be selected, sign(*) indicates the sign function, ibk indicates the current of bridge arm k, N indicates the number of AC-AC conversion submodules in each bridge arm, V c-kj represents the capacitor voltage of the jth AC-AC conversion submodule in the first bridge arm, V con is the bridge arm voltage reference.

5. The distributed finite control set model predictive control method according to claim 1, characterized in that: The basic cost function is the square of the error between the predicted output value of the first bridge arm and the reference output value.

6. The distributed finite control set model predictive control method according to claim 1, characterized in that: The method further comprises: Using a Kalman filter to detect any of the AC-AC conversion submodules to obtain a priori estimates and measurements of the AC-AC conversion submodule; Constructing a single-layer neural network, including configuring network parameters, wherein the network parameters at least include the number of nodes in the input layer, the hidden layer, and the output layer; Using the a priori estimate, the integral of the a priori estimate, the measured value and the integral of the measured value as inputs of the single-layer neural network, and judging whether the modular multi-level matrix converter is abnormal based on the output of the single-layer neural network; The output of the hidden layer is expressed by the following formula: ； The output of the output layer is expressed by the following formula: ； In the formula, represents the activation function, represents the i-th input node of the system, represents the input bias, J and K represent the number of input nodes and hidden nodes respectively, represents the output of the jth hidden layer neuron.

7. The distributed finite control set model predictive control method according to claim 6, characterized in that: The method further comprises: Using a random weight method, obtaining a preset number of weights from an input layer to a hidden layer in the single-layer neural network, and a bias corresponding to the weight; The validity function is used to evaluate the weights obtained each time, and the optimal weights and corresponding biases are selected therefrom.

8. The distributed finite control set model predictive control method according to claim 7, characterized in that: The validity function is expressed by the following formula: ； ； in, represents the validity function, which is used to evaluate the random input weights, is the intermediate variable, represents the error of the kth neuron, and M are learning parameters, and , the superscript T indicates transpose.

9. The distributed finite control set model predictive control method according to claim 7, characterized in that: The method further comprises: Through forward propagation calculation, the weights of the output layer are optimized by minimizing the following objective function, which is expressed by the following formula: ； In the formula, represents the output layer weight, represents the weight corresponding to the jth neuron of the i-th sampling, Represents the output of the i-th sampling.

10. A controller for a modular multi-level matrix converter, characterized in that: The controller adopts the distributed finite control set model predictive control method as described in any one of claims 1 to 9.

Images