Maximum reliability control method for discrete linear non-Gaussian random system
By establishing a state feedback model and optimization algorithm in a discrete linear non-Gaussian stochastic system, the problems of high complexity in non-Gaussian noise processing and unclear performance indicators in existing technologies are solved, achieving efficient and reliable control of the system state within a preset range and improving control performance.
Patent Information
- Application Number
- CN202511914727.5
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-12-18
- Publication Date
- 2026-01-16
AI Technical Summary
Existing non-Gaussian random distribution control methods are difficult to efficiently handle non-Gaussian noise, have high computational complexity and unclear performance indicators, and are difficult to meet the reliability requirements of discrete linear non-Gaussian random systems.
A maximum reliability control method for discrete linear non-Gaussian stochastic systems is proposed. By establishing a state feedback model, deriving the probability density function prediction equation, constructing the Taylor approximation and recursive algorithm optimization problem, and solving the state feedback gain increment, the maximum reliability control of the system state within a preset interval threshold is achieved.
It achieves a simplified calculation process, improves the control performance of non-Gaussian stochastic systems, provides explicit performance guarantees and efficient reliability indicators, and ensures that the system state falls within a preset range with a high probability.
Smart Images

Figure CN121348783A_ABST
Abstract
Description
Technical Field
[0001] This invention belongs to the field of stochastic distribution control technology, specifically relating to a maximum reliability control method for discrete linear non-Gaussian stochastic systems. Background Technology
[0002] Uncertainty is unavoidable in almost all engineering systems, stemming from factors such as parameter variations, fluctuations in operating conditions, and modeling and simulation errors. In the field of control systems, feedback is considered a powerful tool for addressing uncertainty, which has spurred the development of robust and stochastic control methods for different types of uncertainty, such as norm-bounded uncertainty, polyhedral uncertainty, and stochastic uncertainty. Among these, stochastic control has been widely studied and applied due to its ability to effectively handle stochastic uncertainties arising from noise, parameters, and system behavior. Typical stochastic control methods include linear quadratic Gaussian control, minimum variance control, and predictive stochastic control. These traditional methods are mostly based on the Gaussian assumption, thus yielding relatively simple analytical solutions. However, in practical control systems, many random variables do not follow a Gaussian distribution; even if the input is Gaussian, for systems with nonlinear characteristics, the output often exhibits a non-Gaussian distribution, due to the propagation effect of nonlinear uncertainty.
[0003] Non-Gaussian outputs cannot be adequately characterized by mean and variance alone, necessitating the introduction of probability density functions. Against this backdrop, Stochastic Distribution Control (SDC) has attracted widespread attention. SDC is a control framework that describes the system state distribution using probability density functions and is optimized by tracking the desired state distribution or designing statistical indices. In control frameworks optimized by designing statistical indices, the existing Minimum Entropy Control (MEC) method achieves output distribution convergence by minimizing entropy, effectively addressing non-Gaussian random noise in the system. However, it suffers from high computational complexity, difficulty in using it as an independent performance indicator, and unclear performance boundaries. Therefore, there is an urgent need for a statistical index with clear probabilistic meaning, low computational complexity, and explicit performance guarantees to achieve more efficient non-Gaussian stochastic distribution control.
[0004] Probability is the most readily conceivable alternative criterion. When calculating probability, the integration region of the probability density function needs to be determined. Under a stochastic distribution control framework, the probability density function information can be derived at each sampling time, and its integration region can be selected as a preset interval threshold. Since the reliability index in reliability describes the probability that the system will achieve a preset function or performance under specific time and conditions, it can serve as a necessary probability-based alternative metric. Common reliability modeling methods also achieve this by comparing the relationship between stochastic dynamic processes and corresponding performance thresholds. Therefore, this patent proposes a new reliability criterion for non-Gaussian stochastic distribution control to measure the probability that the system state will remain within a preset interval threshold. By using reliability as the optimization objective, non-Gaussian stochastic distribution control based on the reliability criterion can be achieved, and this patent first focuses on related research for discrete linear non-Gaussian stochastic systems. Summary of the Invention
[0005] To address the problem that existing non-Gaussian random distribution control methods still struggle to efficiently handle non-Gaussian random noise, this invention proposes a maximum reliability control method for discrete linear non-Gaussian random systems. Focusing on the non-Gaussian noise control problem of discrete linear non-Gaussian random systems, and based on the idea of maximizing system reliability, this invention proposes a method that can control the system state under the influence of non-Gaussian noise within a preset threshold range with the greatest probability. The feasibility of the method is verified through case studies, providing a new index design approach for improving the control performance of general non-Gaussian random systems.
[0006] To achieve the above objectives, the present invention adopts the following technical solution:
[0007] A method for maximizing reliability control of discrete linear non-Gaussian stochastic systems includes the following steps:
[0008] Step 1: Establish a discrete linear non-Gaussian stochastic system model and clarify the maximum reliability control problem with state feedback.
[0009] Step 2: Derive the one-step probability density function prediction equation for the closed-loop control system;
[0010] Step 3: Establish a one-step reliability prediction model for the closed-loop control system;
[0011] Step 4: Construct the maximum reliability control optimization problem using Taylor approximation and recursive algorithms;
[0012] Step 5: Solve the optimization problem to obtain the vectorized state feedback gain increment, and obtain the desired state feedback gain through matrix reconstruction.
[0013] Specifically, in step 1, the discrete linear non-Gaussian stochastic system model is established as follows:
[0014] (1); In the formula, the subscript k represents the k-th time point. express The system state vector with dimension n at time step This represents a control input vector of dimension m. Let A represent a random vector of process noise with dimension n, and let A, B, and H be matrices of appropriate dimensions.
[0015] Identify the maximum reliability control problem with state feedback: design the optimal dimension for each step as follows. State feedback gain matrix This makes the control law Under the influence of the system, the system state (error) can fall within the preset threshold range with the maximum achievable reliability (probability). Inside, among which As a preset upper limit, each element is greater than 0. For preset lower limit and .For example, The following design is possible:
[0016] ; In the formula, and These represent the final value and maximum value of the preset threshold function, respectively. and To adjust the coefficients of the convergence rate, It represents the convergence time, which can also be understood as the boundary between the transient process and the steady-state process.
[0017] Specifically, in step 2, the one-step probability density function prediction equation of the closed-loop control system is derived as follows:
[0018] (2); In the formula, express System state vector at time step The probability density function, express The specific implementation value, express The probability density function, .
[0019] Specifically, in step 3, the one-step reliability prediction model of the closed-loop control system is established as follows:
[0020] (3); In the formula, express System state vector at time step Falling within the preset range threshold The reliability (probability) within the range is determined by the one-step probability density function prediction equation obtained in step 2 within a preset threshold interval. The result is obtained from the integration on the 1st.
[0021] Specifically, in step 4, the maximum reliability control optimization problem is constructed using Taylor approximation and recursive algorithms as follows: (4); In the formula, This means maximizing the objective function. Solving for the increment of the vectorized state feedback gain , , and They represent and State feedback gain after time vectorization and They represent and Time-state feedback gain matrix and Elements in; , , Indicates non-negative weighting coefficients; Indicates control of energy. The dimension is and state The relevant matrix, for a specific system, can be uniquely determined; This represents the sum of the squares of each element of the state feedback gain increment, used to avoid excessive changes in gain. The approximation of the reliability prediction model in step 3 using the second Taylor approximation can be expressed as follows: ,in, , , , Indicates the sign of partial derivative. Indicates AND The relevant functions are known The value at that location.
[0022] Specifically, in step 5, the vectorized state feedback gain increment is calculated as follows:
[0023] (5); In the formula, , Representation matrix The reverse, Indicates size is The identity matrix. Solve for... Afterwards, through And on The desired state feedback gain can be obtained by performing matrix reconstruction. .
[0024] Beneficial effects:
[0025] This invention addresses the problem that existing non-Gaussian stochastic distribution control methods still struggle to efficiently handle non-Gaussian noise. It proposes a maximum reliability control method for discrete linear non-Gaussian stochastic systems, starting from the basic form of state feedback control and using Taylor approximation and recursive algorithms to achieve online solution of the maximum reliability control optimization problem. With its simple principles and computational process, this invention provides a new approach to index design for improving the control performance of general non-Gaussian stochastic systems. Attached Figure Description
[0026] Figure 1 This is a flowchart of a maximum reliability control method for a discrete linear non-Gaussian stochastic system according to the present invention;
[0027] Figure 2 The system state response is obtained by using the proposed maximum reliability control method and fixed gain method; where (a) is the maximum reliability control method and (b) is the fixed gain method. Detailed Implementation
[0028] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention. Furthermore, the technical features involved in the various embodiments of this invention described below can be combined with each other as long as they do not conflict with each other.
[0029] like Figure 1 As shown, the present invention provides a maximum reliability control method for discrete linear non-Gaussian stochastic systems, the method comprising the following steps:
[0030] Step 1: Establish a discrete linear non-Gaussian stochastic system model and clarify the maximum reliability control problem with state feedback.
[0031] Step 2: Derive the one-step probability density function prediction equation for the closed-loop control system;
[0032] Step 3: Establish a one-step reliability prediction model for the closed-loop control system;
[0033] Step 4: Construct the maximum reliability control optimization problem using Taylor approximation and recursive algorithms;
[0034] Step 5: Solve the optimization problem to obtain the increment of the vectorized state feedback gain, and obtain the desired state feedback gain through matrix reconstruction.
[0035] Specifically, in step 1, the discrete linear non-Gaussian stochastic system model is established as follows:
[0036] (1), where the subscript k represents the k-th time point, express The system state vector with dimension n at time step This represents a control input vector of dimension m. Let A represent a random vector of process noise with dimension n, and let A, B, and H be matrices of appropriate dimensions.
[0037] Identify the maximum reliability control problem with state feedback: design the optimal dimension for each step as follows. State feedback gain matrix This makes the control law Under the influence of the system, the system state (error) can fall within the preset threshold range with the maximum achievable reliability (probability). Inside, among which As a preset upper limit, each element is greater than 0. For preset lower limit and .For example, The following design is possible:
[0038] ; In the formula, and These represent the final value and maximum value of the preset threshold function, respectively. and To adjust the coefficients of the convergence rate, It represents the convergence time, which can also be understood as the boundary between the transient process and the steady-state process.
[0039] Specifically, in step 2, the one-step probability density function prediction equation of the closed-loop control system is derived as follows:
[0040] (2); In the formula, express System state vector at time step The probability density function, express The specific implementation value, express The probability density function, .
[0041] Specifically, in step 3, the one-step reliability prediction model of the closed-loop control system is established as follows:
[0042] (3); In the formula, express System state vector at time step Falling within the preset range threshold The reliability (probability) within the range is determined by the one-step probability density function prediction equation obtained in step 2 within a preset threshold interval. The result is obtained from the integration on the 1st.
[0043] Specifically, in step 4, the maximum reliability control optimization problem is constructed using Taylor approximation and recursive algorithms as follows:
[0044] (4); In the formula, This means maximizing the objective function. Solving for the increment of the vectorized state feedback gain , , and They represent and State feedback gain after time vectorization and They represent and Time-state feedback gain matrix and Elements in; , , Indicates non-negative weighting coefficients; Indicates control of energy. The dimension is and state The relevant matrix, for a specific system, can be uniquely determined; This represents the sum of the squares of each element of the state feedback gain increment, used to avoid excessive changes in gain. The approximation of the reliability prediction model in step 3 using the second Taylor approximation can be expressed as: ;
[0045] in, , , , Indicates the sign of partial derivative. Indicates AND The relevant functions are known The value at that location.
[0046] Specifically, in step 5, the vectorized state feedback gain increment is calculated as follows:
[0047] (5); In the formula, , Representation matrix The reverse, Indicates size is The identity matrix. Solve for... Afterwards, through And on The desired state feedback gain can be obtained by performing matrix reconstruction. .
[0048] Just as minimum entropy control requires additional conditions to ensure system stability, maximum reliability control provides the following stability check and adjustment conditions after each solution step to ensure stability: 1) Select the initial state feedback gain to make the matrix The 2-norm is less than 1, that is 2) During the state feedback gain update process, if the following occurs... In this case, the newly updated optimal feedback gain can be reset to the value of the previous time step, or a small adjustment can be made near the optimal solution. In optimal control theory, stability is often given less attention because, under fairly general conditions, the system is usually stable. Based on this experience, events... The probability of this happening is extremely low.
[0049] Example:
[0050] Step 1, establish the discrete linear non-Gaussian stochastic system model as follows:
[0051] (6); In the formula, the initial state is set to , It follows the distribution as follows:
[0052] (7); Identify the maximum reliability control problem with state feedback: design the optimal dimension for each step as follows. State feedback gain matrix This makes the control law Under the influence of the system, the system state (error) can fall within the preset threshold range with the maximum achievable reliability (probability). Inside, among which As a preset upper limit, each element is greater than 0. For preset lower limit and In this embodiment, and The following design is possible:
[0053] ; In the formula, , , , .
[0054] Step 2, derive the one-step probability density function prediction equation for the closed-loop control system as follows:
[0055] (8); In the formula, express System state vector at time step The probability density function, express The specific implementation value, express The probability density function, .
[0056] Step 3: Establish the one-step reliability prediction model for the closed-loop control system as follows:
[0057] (9); In the formula, express System state vector at time step Falling within the preset range threshold The reliability (probability) within the range is determined by the one-step probability density function prediction equation obtained in step 2 within a preset threshold interval. The result is obtained from the integration on the 1st.
[0058] Step 4: Construct the maximum reliability control optimization problem using Taylor approximation and recursive algorithms as follows:
[0059] (10); In the formula, , , .
[0060] Step 5, calculate the vectorized state feedback gain increment as follows:
[0061] (11); In the formula, , Representation matrix The reverse, Indicates size is The identity matrix. Solve for... Afterwards, through And on The desired state feedback gain can be obtained by performing matrix reconstruction. .
[0062] The system state response obtained by the maximum reliability control method proposed in this embodiment is compared with the system state response obtained by the fixed gain method. Figure 2 This is the comparison result of the system state response.
[0063] from Figure 2 It can be seen that the system state response obtained by the proposed maximum reliability control method can meet the preset interval threshold well, while the system state response obtained by the fixed gain method significantly exceeds the preset interval threshold. When the gain is 0.1, the proposed method can control the system state response within the specified performance threshold with at least 96% probability, while the given fixed gain can only reach 0.3%.
[0064] The present invention also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the steps of the above-described maximum reliability control method for discrete linear non-Gaussian stochastic systems based on reliability criteria.
[0065] The present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the above-described maximum reliability control method for discrete linear non-Gaussian stochastic systems based on reliability criteria.
[0066] The contents not described in detail in this specification are existing technologies known to those skilled in the art.
[0067] Those skilled in the art will understand that embodiments of the present invention can be provided as methods or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code. The solutions in the embodiments of the present invention can be implemented using various computer languages, such as the object-oriented programming language Java and the interpreted scripting language JavaScript.
[0068] This invention is described with reference to flowchart illustrations and / or block diagrams of methods and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.
[0069] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.
[0070] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.
[0071] Although preferred embodiments of the invention have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including both the preferred embodiments and all changes and modifications falling within the scope of the invention.
[0072] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.
Claims
1. A maximum reliability control method for discrete linear non-Gaussian stochastic systems, characterized in that, Includes the following steps: Step 1: Establish a discrete linear non-Gaussian stochastic system model and clarify the maximum reliability control problem with state feedback. Step 2: Derive the one-step probability density function prediction equation for the closed-loop control system; Step 3: Establish a one-step reliability prediction model for the closed-loop control system; Step 4: Construct the maximum reliability control optimization problem using Taylor approximation and recursive algorithms; Step 5: Solve the optimization problem to obtain the vectorized state feedback gain increment, and obtain the desired state feedback gain through matrix reconstruction.
2. The maximum reliability control method for a discrete linear non-Gaussian stochastic system according to claim 1, characterized in that, In step 1, the discrete linear non-Gaussian stochastic system model is established as follows: (1); In the formula, the subscript k represents the k-th time point. express The system state vector with dimension n at time step This represents a control input vector of dimension m. Let A, B, and H represent the noise vector of a non-Gaussian random process with dimension n, where A, B, and H are matrices of appropriate dimensions. Identify the maximum reliability control problem with state feedback: design the optimal dimension for each step as follows. State feedback gain matrix This makes the control law Under the influence of the system, the system state can fall within the preset threshold range with the maximum achievable reliability. Inside, among which As a preset upper limit, each element is greater than 0. For preset lower limit and .
3. The maximum reliability control method for a discrete linear non-Gaussian stochastic system according to claim 1, characterized in that, In step 2, the one-step probability density function prediction equation for the closed-loop control system is derived as follows: (2); In the formula, express System state vector at time step The probability density function, express The specific implementation value, express The probability density function, .
4. The maximum reliability control method for a discrete linear non-Gaussian stochastic system according to claim 1, characterized in that, In step 3, the one-step reliability prediction model for the closed-loop control system is established as follows: (3); In the formula, express System state vector at time step Falling within the preset range threshold The reliability within the range is determined by the one-step probability density function prediction equation obtained in step 2 within a preset threshold interval. The result is obtained from the integration on the 1st.
5. The maximum reliability control method for a discrete linear non-Gaussian stochastic system according to claim 1, characterized in that, In step 4, the maximum reliability control optimization problem is constructed using Taylor approximation and recursive algorithms as follows: (4); In the formula, This means maximizing the objective function. Solving for the increment of the vectorized state feedback gain , , and They represent and The state feedback gain after time vectorization. and They represent and Time-state feedback gain matrix and Elements in; , , Indicates non-negative weighting coefficients; Indicates control of energy. Indicates dimension as and state The relevant matrix, for a specific system, can be uniquely determined; This represents the sum of the squares of each element of the state feedback gain increment, used to avoid excessive changes in gain. The approximation of the reliability prediction model in step 3 using the second Taylor approximation can be expressed as follows: ,in, , , , Indicates the sign of partial derivative. Indicates AND The relevant functions are known The value at that location.
6. The maximum reliability control method for a discrete linear non-Gaussian stochastic system according to claim 1, characterized in that, In step 5, the vectorized state feedback gain increment is calculated as follows: (5); In the formula, , Representation matrix The reverse, Indicates size is The identity matrix; solve Afterwards, through And on The desired state feedback gain can be obtained by performing matrix reconstruction. .
