Intelligent warehouse temperature regulation and control method based on distributed monitoring

Abstract

The invention relates to an intelligent warehouse temperature regulation and control method based on distributed monitoring, and the method comprises the steps: firstly constructing an intelligent warehouse temperature control system model; then, combining an output equivalent transformation and detectability decomposition technology, designing a self-adaptive distributed observer based on a consistency theory, and realizing collaborative estimation of the warehouse temperature; aiming at system measurement output, an interval observer is further constructed, an algebraic relationship between an output vector and unknown input is extracted, an unknown input algebraic reconstruction mechanism decoupled from control input is further established, and accurate valuation of the unknown input is realized; and finally, combining the temperature state estimation value and an unknown input estimation value, and combining a weighted averaging method and a feedback control strategy to design an intelligent warehouse temperature regulation controller. According to the technical scheme, the fault tolerance and robustness of the intelligent warehouse temperature monitoring and regulation system are effectively improved, and the method has good engineering application value and popularization prospects.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of warehouse temperature regulation, and particularly relates to an intelligent warehouse temperature regulation method based on distributed monitoring. BACKGROUND

[0002] With the rapid development of wireless communication technology and sensor technology, distributed state estimation technology based on wireless sensor networks has become a key support means in intelligent sensing and regulation systems. This technology builds a multi-sensor node cooperative sensing network, so that each node only needs to obtain local information, and the global state is accurately reconstructed by means of distributed estimation algorithm, thereby reducing the dependence on the accuracy of single-point sensors and improving the robustness and fault tolerance of the system.

[0003] The application field of distributed observation technology includes the field of intelligent warehouse temperature monitoring. By deploying a large number of low-cost temperature sensor nodes in the warehouse, a wireless cooperative sensing network is constructed, which not only breaks through the limitations of traditional temperature monitoring systems relying on a single measuring point, but also realizes high-resolution dynamic modeling of the thermal distribution of the warehouse environment. Each sensor node measures locally and exchanges information with neighboring nodes, and uses distributed state estimation methods to cooperatively reconstruct the overall temperature field, which can effectively cope with uncertain factors such as local sensor failure and communication delay, and significantly improve the stability and response capability of the environmental monitoring system. This technology provides a solid technical support for ensuring the quality of warehouse goods temperature and humidity environment, preventing heat accumulation risks, and realizing green energy-saving regulation, which meets the development needs of intelligent logistics and green storage.

[0004] However, existing distributed observation technology mainly focuses on eliminating the negative effects of unknown inputs and cooperatively estimating system states, such as documents [1] and [2]: Document [1]: Distributed unknown input observer, IEEE Transactions on Automatic Control, vol. 68, no. 12, pp. 8244-8251, 2023. Document [1] proposes a distributed unknown input observer design scheme based on detectability decomposition, which is used to solve the problem of distributed estimation of the state of a linear system with unknown inputs.

[0005] Document [2]: Adaptive distributed unknown input observer for linear systems, Applied Mathematics and Computation, vol. 486, p. 129027, 2025. Document [2] establishes an adaptive unknown input observer by combining disturbance decoupling and follower consensus strategy, which realizes the collaborative estimation of system state while avoiding the dependence on global information.

[0006] In fact, unknown inputs not only exist widely in actuator channels, but also exist universally in sensor channels. The unknown inputs in actuator channels usually include actuator faults and external disturbances, while the unknown inputs in sensor channels usually include sensor faults and measurement noises. For the intelligent warehouse temperature monitoring and regulation system, existing documents [1] and [2] mainly focus on suppressing the interference of unknown inputs on system performance, but ignore the estimation and compensation control of unknown inputs themselves, which is difficult to fundamentally improve the stability and regulation accuracy of the system. Therefore, it is urgent to research and develop a new monitoring and regulation strategy that integrates unknown input estimation and compensation mechanism. In addition, existing documents [1] and [2] both assume that the control input of the system is completely visible to all sensor nodes. However, in actual applications, each sensor node can only access a part of the control input vector, i.e. each node can only obtain information of some input components. This locality of information acquisition further aggravates the complexity of collaborative state estimation and feedback control, and also puts higher requirements on the adaptability and generalizability of existing methods. SUMMARY

[0007] The present application provides an intelligent warehouse temperature regulation method based on distributed monitoring to overcome the shortcomings of the prior art.

[0008] The present application is realized by the following technical solutions: An intelligent warehouse temperature regulation method based on distributed monitoring, comprising the following steps: S1: Based on the heat exchange mechanism between the warehouse partitions and the outside world, combined with the temperature measurement data collected by the sensors uniformly distributed in the warehouse, an intelligent warehouse temperature control system model with unknown inputs is constructed; S2: According to the intelligent warehouse temperature control system model, a distributed observer based on consensus theory is constructed using detectability decoupling technology, and the collaborative estimation of the temperature of each partition is obtained to obtain the temperature estimate of the intelligent warehouse; S3: To achieve compensatory control of unknown inputs, an interval observer is constructed for the output variables of the smart warehouse temperature control system model to obtain the algebraic relationship between the unknown input and output variables. Based on the estimated value of the smart warehouse temperature, an algebraic reconstruction mechanism for the unknown input is established to obtain the estimated value of the unknown input. S4: Based on the estimated value of the smart warehouse temperature and the estimated value of the unknown input, and combined with the weighted mean method and feedback control strategy, a smart warehouse temperature control controller is designed to maintain the temperature of the smart warehouse at the expected value in the presence of unknown input.

[0009] Furthermore, in step S1, the intelligent warehouse temperature control system model is: (1) in, Indicates the temperature of the smart warehouse, Indicates the first The temperature of each zone, , is the number of partitions of the smart warehouse, ; represents the control input, Indicates the control input A quantity, , is a positive integer; Represents unknown input, used to describe external environmental disturbances or unmodeled heat source effects. is a positive integer; represents the heat transfer coefficient matrix, , Indicates the Partition and The heat exchange coefficient between the partitions, and represents a constant matrix; represents the temperature measurement output vector collected by sensors evenly spaced throughout the warehouse; Indicates the The measurement output vector collected by the sensors, , is the number of sensors; , is a positive integer; , Indicates the The measurement noise in each sensor channel, , is a positive integer; accordingly, the constant matrix and It can be expressed in blocks as and , therefore, for the sensors, whose measured output vector It can be expressed as: (2) in and represents a constant matrix; Strongly connected graph is used as the communication topology between sensor nodes Indicates that represents the set of sensor nodes, represents the sensor edge set, represents the sensor adjacency weight matrix, if A sensor can obtain the The information of each sensor, then the connection weight , if the The sensor cannot obtain The information of each sensor, then the connection weight On Strongly Connected Graphs The Laplace matrix of Defined as ,in , .

[0010] Furthermore, since each sensor node can only access part of the control input of the intelligent warehouse temperature control system model, sensors, It can be expressed as: (3) in and Represents control input the accessible and inaccessible parts of and is a positive integer, satisfying , and is a constant matrix; Due to unknown input It is unknown for all sensors, so for the sensors, lumped unknown input and its coefficient matrix It can be described as: (4) For the Sensors, combined with equations (1) to (4), the improved intelligent warehouse temperature control system model is: (5) The coefficient matrix in equation (5) satisfies the rank condition as follows: (6) where represents a unit matrix of dimension .

[0011] Further, step S2 includes the following sub-steps: S2.1: for the improved intelligent warehouse temperature control system model, in order to eliminate the negative impact of the , an equivalent output transformation is introduced, where represents a unit matrix of dimension , represents the generalized inverse matrix of matrix ; then, the improved intelligent warehouse temperature control system model is described as: (7) where ; The matrix is defined as , the matrix and satisfy , where represents the undetectable subspace of the matrix pair ; Lemma 1: ; Lemma 1 shows that the following matrix equation has a solution, (8) The general form of the solution matrix is , where is an arbitrary constant matrix; Lemma 2: define , then the undetectable subspaces of the matrix pair and are the same, that is ; S2.2: perform detectability decoupling on equation (7); let represent the dimension of the undetectable subspace ; define the matrix , whose column vectors consist of a set of orthogonal bases of the undetectable subspace ; at the same time, define the matrix , whose column vectors consist of a set of orthogonal bases of the undetectable subspace It is composed of a set of standard orthogonal bases of the orthogonal complement space of ; from this we can see that ,in Representation matrix The image space, Represents the matrix nuclear space; To obtain the matrix pair Decoupling the detectability of , constructing an orthogonal transformation matrix ; Due to the undetectable subspace It's about Invariant, then in the orthogonal transformation matrix Under the action of and Detectability breakdown: (9) in , , , ; Matrix pair Is detectable, matrix It is unstable; S2.3: According to the equivalent output transformation in step S2.1 and the detectability decomposition in step S2.2, for equation (7), the distributed observer is: (10) in, Indicates the The state of the observer, express estimated value of; Constant Matrix for: ; Feedback gain matrix for: , where the matrix Make is a Hurwitz matrix; Gain Matrix for: ; Coupling gain Designed for ,in is the adaptive law, The update rules are as follows: (11) (12) represents the initial moment, and are positive design parameters.

[0012] Further, step S3 comprises: S3.1: due to the unknown input belongs to the actuator channel, i.e. , there exists a matrix such that ; accordingly, the matrix can be expressed as ; moreover, in order to realize the reconstruction of the lumped unknown input , there exists at least one sensor node in the multi-sensor network such that the matrix is column full rank; without loss of generality, it is assumed that there exist sensor nodes satisfying that the matrix is column full rank; according to equation (7), the time derivative of the equivalent output variable is obtained as: (35) The interval observer for the equivalent output variable is: (36) where and denote the upper bound state and the lower bound state of the interval observer, whose initial conditions satisfy and , and denote the upper bound and the lower bound of the system initial condition , satisfying , and denote the upper bound and the lower bound of the lumped unknown input , satisfying , the matrix is a constant matrix satisfying both Hurwitz matrix condition and Metzler matrix condition; there exists a time constant such that when , the states and of the interval observer satisfy ; According to equation (36), there exists a time-varying function satisfying: (41) where and for any , each ranging from 0 to 1 ; According to equation (41), the function can be expressed as: (42) where if , then if , then ; Combining equation (36) and equation (41), the time derivative of the equivalent output variable can be described as: (43) where, (44) Rearranging equation (35) and equation (43), an algebraic expression about the lumped unknown input is obtained: (45) Step 3.2: Since in equation (43) is unknown, an estimation of is provided by a sliding mode differentiator as follows: (46) where and denote the design parameters of the sliding mode differentiator, and are the first and second states of the sliding mode differentiator, respectively, will asymptotically converge to in finite time, i.e. ; For the lumped unknown input in equation (7), an algebraic reconstruction mechanism of the lumped unknown input is established as follows according to equation (45), the estimation of provided by equation (46), and the estimation of the smart warehouse temperature provided by the distributed observer in Step 2: (47) where denotes the reconstructed value of the lumped unknown input ; For the unknown input , the algebraic reconstruction mechanism of the unknown input is established as follows according to equation (4) and equation (47): (48) where​ Indicates unknown input The reconstructed value of .

[0013] Furthermore, in step S4, the intelligent warehouse temperature adjustment controller is: (51) in is the feedback gain matrix, which is selected to satisfy is the Hurwitz matrix, represents the expected value of temperature, which is given by the reference dynamic system whose state evolution satisfies ,in Indicates external input signal.

[0014] The beneficial effects that can be achieved by the present invention are: (1) A distributed temperature observation scheme for intelligent warehouses based on a multi-sensor network is proposed. By unifying the control input components that cannot be obtained by sensor nodes and the unknown inputs in the actuator channels as lumped unknown inputs, and combining detectability decoupling with output transformation methods, an adaptive distributed observer is constructed. This observer effectively eliminates the adverse effects of unknown inputs and unmeasurable control input components in the actuator and sensor channels on the system observation performance, and achieves high-precision collaborative estimation of the temperature field in the intelligent warehouse. At the same time, this method does not require the introduction of additional state variables or the expansion of system dimensions during the distributed observer design process, and has high engineering feasibility and computational efficiency.

[0015] (2) An interval observer is designed to establish the algebraic relationship between unknown inputs and system outputs, and based on this, an unknown input reconstruction mechanism is proposed. This mechanism can not only achieve accurate estimation of unknown inputs, but also decouple the structure from the adjustment input of the intelligent warehouse temperature control system, ensuring that the reconstructed unknown input information can be directly used in controller design, thereby achieving compensation control for unknown inputs. Compared with the methods in references [1] and [2] that are limited to suppressing the influence of unknown inputs and lack effective estimation, this invention breaks through their limitations in system uncertainty compensation. This method breaks through their limitations in system uncertainty compensation and significantly improves the robustness and anti-interference ability of the intelligent warehouse temperature monitoring and control system to external disturbances and unknown inputs. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 This is a flow chart of the intelligent warehouse temperature control method based on distributed monitoring of the present invention.

[0017] Figure 2 Schematic diagram of intelligent warehouse partitioning and measurement information in an embodiment of the present invention.

[0018] Figure 3 This is a sensor network communication topology diagram in an embodiment of the present invention.

[0019] Figure 4 Schematic diagram of the second norm of temperature estimation error in the smart warehouse in an embodiment of the present invention.

[0020] Figure 5 Schematic diagram of the adaptive law in an embodiment of the present invention.

[0021] Figure 6 Schematic diagram of unknown input reconstruction of the intelligent warehouse in an embodiment of the present invention.

[0022] Figure 7 Schematic diagram of temperature changes in each zone of the smart warehouse in an embodiment of the present invention. DETAILED DESCRIPTION

[0023] The present invention is described in detail below with reference to the accompanying drawings and specific embodiments. This embodiment is implemented based on the technical solution of the present invention, and provides a detailed implementation method and specific operation process, but the protection scope of the present invention is not limited to the following embodiments.

[0024] Specifically, such as Figure 1 As shown, the present invention provides a temperature control method for an intelligent warehouse based on distributed monitoring, which includes the following steps: Step 1: Based on the heat exchange mechanism between warehouse partitions and the outside world, combined with the temperature measurement data collected by sensors (temperature sensors, referred to as sensors) evenly distributed in the warehouse, a smart warehouse temperature control system model with unknown inputs is established.

[0025] The steps include: Step 1.1, for To design a smart warehouse with multiple partitions, we first obtain the warehouse's structural parameters, including the spatial dimensions of each partition, the thermal properties of the wall materials, and the area of ​​the partition partitions. Based on these structural parameters and incorporating Fourier's law of heat conduction and Newton's law of cooling, we analyze the heat conduction and convection between the warehouse's partitions, as well as the heat exchange paths with the external environment. This allows us to determine the heat exchange relationships between adjacent partitions and between a partition and the outside world. Combined with the partition temperature measurement data collected by sensors, we establish a smart warehouse temperature control system model, whose state equation can be expressed as: (1) in, Indicates the temperature of the smart warehouse, Indicates the first The temperature of each zone, , represents the control input, Indicates the control input A quantity, , is a positive integer; Represents unknown input, used to describe external environmental disturbances or unmodeled heat source effects. is a positive integer; Represents the heat transfer coefficient matrix , Indicates the Partition and The heat exchange coefficient between the partitions, and represents a constant matrix. represents the temperature measurement output vector collected by sensors evenly spaced throughout the warehouse, represents an unknown input in a sensor channel, Indicates the The measurement output vector collected by the sensors, , is the number of sensors; , is a positive integer; Indicates the The measurement noise in each sensor channel, , is a positive integer; accordingly, the constant matrix and It can be expressed in blocks as and , therefore, for the sensors, whose measured output vector It can be expressed as: (2) in and represents a constant matrix.

[0026] Strongly connected graph is used as the communication topology between sensor nodes Indicates that represents the set of sensor nodes, represents the sensor edge set, represents the sensor adjacency weight matrix, if A sensor can obtain the The information of each sensor, then the connection weight , if the The sensor cannot obtain the The information of each sensor, then the connection weight On Strongly Connected Graphs The Laplace matrix of Defined as where , .

[0027] Step 1.2, based on the smart warehouse temperature control system model (1) established in step 1.1, decompose the control input for the i-th sensor, and get an improved system model suitable for the design of a distributed observer. Note that each sensor node can only access part of the control input of the smart warehouse temperature control system model. Therefore, for the i-th sensor, can be expressed as: (3) where and represent the available and unavailable parts of the control input , and are positive integers satisfying , and are constant matrices. Note that the unknown input is unknown to all sensors, so for the i-th sensor, the aggregated unknown input and its coefficient matrix can be described as (4) Combining equations (1)-(4) for the i-th sensor, the improved smart warehouse temperature control system model can be expressed as (5) The coefficient matrix in equation (5) satisfies the following rank condition: (6) where represents an identity matrix of dimension , .

[0028] Step 2, based on the smart warehouse temperature control system model established in step 1, design a distributed observer to cooperatively estimate the temperature of the smart warehouse.

[0029] includes the following steps: Step 2.1, considering the improved smart warehouse temperature control system model (5) established in step 1.2, in order to eliminate the negative effects of the unknown input in the sensor channel, an equivalent output transformation is introduced, where represents​​​​ identity matrix, denotes the generalized inverse matrix of the matrix . Then, the improved intelligent warehouse temperature control system model (5) can be re-described as (7) where .

[0030] Define the matrix as , the matrix and satisfy , where denotes the undetectable subspace of the matrix pair .

[0031] Lemma 1: .

[0032] Proof: According to the matrix rank condition (6), we have

[0033] This shows that . The proof is complete.

[0034] In addition, Lemma 1 shows that the following matrix equation has a solution, (8) whose solution matrix is generally of the form , where is an arbitrary constant matrix.

[0035] Lemma 2 [1]: Define , then the undetectable subspaces of the matrix pair and are the same, that is,

[0036] Step 2.2, according to the formula (7) of step 2.1, decouple the detectability. Let denote the dimension of the undetectable subspace . Define the matrix whose column vectors consist of a set of orthogonal basis of the undetectable subspace , and define the matrix whose column vectors consist of a set of standard orthogonal basis of the orthogonal complement space of the undetectable subspace . It can be known that , where denotes the image space of the matrix , denotes the orthogonal complement space of the matrix the null space of

[0037] To obtain the detectability decoupling of the matrix pair , a orthogonal transformation matrix is constructed as It is noted that the undetectable subspace is invariant with respect to (i.e. the -invariant subspace). This means that under the action of the orthogonal matrix , the detectability decomposition with respect to the matrix pair and is obtained as (9) where , , , . The matrix pair is detectable, and the matrix is unstable.

[0038] Step 2.3. According to the equivalent output transformation in Step 2.1 and the detectability decomposition in Step 2.2, a distributed observer is designed for the smart warehouse temperature control system model (7) as follows: (10) where denotes the state of the th observer, and denotes the estimate of . The constant matrix is defined as , and the feedback gain matrix is designed as , where the matrix is chosen such that is a Hurwitz matrix, and the gain matrix is chosen as .

[0039] In addition, the coupling gain is designed as , where is an adaptive law, whose update rule is as follows: (11) (12) denotes the initial time , and are positive design parameters.

[0040] The distributed observer implements the verification procedure for the temperature cooperative estimation of the smart warehouse as follows: First, define the temperature estimation error as According to equation (7) and equation (10), we can get (13) where represents the element of the Laplacian matrix in the i-th row and j-th column.

[0041] Using the orthogonal transformation matrix , the temperature estimation error is decoupled into the following form (14) where and represent the detectable part and the undetectable part of the temperature estimation error, respectively. According to equation (13) and (14), we have (15) Let and , then we can get the following compact form (16) where , , , , , , , .

[0042] Define the overall temperature estimation error as , from equation (16) we can get (17) where , and .

[0043] Then, for equation (16), introduce the following transformation (18) Further, we can get (19) where

[0044] According to the transformation (18), (12) can be further represented as​​ (20) By the definition, there exists a positive definite matrix satisfying (21) where denotes the minimum eigenvalue of the matrix , , .

[0045] Select the Lyapunov function as follows (22) where is a sufficiently large constant. Combining (19) and (22), we can get the time derivative of (23) In addition, the time derivative of (24) Let , , and , then according to (11), (19) and (20), we can get (25) (26) At the same time, we can also get the following relationship (27) Substituting (25)-(27) into (24), we get (28) where . Using Young's inequality, we can get (29) Then (28) can be further expressed as (30) Note that (31) Combining (23), (30) and (31), we have (32) where According to equation (32), a sufficiently large can be guaranteed , which indicates that the value of Lyapunov function will not increase. Recall equation (22), it can be found that , and are bounded. In addition, according to equation (19), and are also bounded. Furthermore, according to equation (20), is also bounded. Therefore, is uniformly continuous. Note that and , it can be seen that exists a finite limit, i.e. . Let (33) it can be found that (34) which means exists and is bounded. According to Barbalat's lemma, there is . In addition, it is noted that is positive definite at any time, which indicates that and are asymptotically stable. Since the transformations (14) and (18) are non-singular, it can be concluded that the temperature estimation error asymptotically converges to zero, i.e. .

[0046] Therefore, the distributed observer can cooperatively estimate the temperature of the smart warehouse.

[0047] Step 3, in order to realize the compensation control of unknown input, an interval observer is designed for the equivalent output transformed system output vector to obtain the algebraic relationship between the lumped unknown input and the system output. Thus, the algebraic reconstruction mechanism of the lumped unknown input decoupled from the control input in the smart warehouse temperature control system model is established, the accurate reconstruction of the lumped unknown input is realized, and finally the estimated value of the unknown input in the smart warehouse temperature control system model is obtained.

[0048] comprising the following steps: Step 3.1, an interval observer is established for the output vector in equation (7) to obtain the algebraic relationship between the lumped unknown input and the system output variable, and the specific design process is as follows: Firstly, it is noted that the unknown input in the smart warehouse temperature control system model (1) belongs to the actuator channel, i.e. . Therefore, there exists a matrix such that . Accordingly, the matrix may be expressed as Moreover, in order to realize the reconstruction of the lumped unknown input, there exists at least one sensor node in the multi-sensor network such that the matrix is column full rank. Without loss of generality, assume that there exist sensor nodes satisfying that the matrix is column full rank. According to the expression (7) in Step 2.1, the time derivative of the equivalent output variable is obtained as (35) For the equivalent output variable , the interval observer is established as (36) where and denote the upper bound state and the lower bound state of the interval observer, whose initial conditions satisfy and , and denote the upper bound and the lower bound of the system initial condition , satisfying , and denote the upper bound and the lower bound of the lumped unknown input , satisfying , the matrix is a constant matrix satisfying both the Hurwitz matrix condition and the Metzler matrix condition. There exists a time constant such that when , the states and of the interval observer satisfy .

[0049] The interval observer realizes the interval estimation of the output variable in the intelligent warehouse temperature control system model. The verification process is as follows: Define the interval estimation error as and , combining the expressions (35) and (36), we can obtain (37) It can be seen from that (38) According to the distributed observer designed in Step 2, the temperature estimation error is asymptotically stable. This indicates that there exists a time constant such that (39) holds for any . Moreover, it is noted that the initial condition of the interval estimation error system (37) satisfies (40) Meanwhile, is both a Hurwitz matrix and a Metzler matrix, it is known from the theory of positive systems that the interval estimation error system (37) is non-negative when . This further implies that for any , there holds .

[0050] Therefore, the interval observer can provide an interval estimation of the equivalent system output vector.

[0051] From the interval observer (36), there exists a time-varying function satisfying (41) where and , for any , are both in the range of 0 to 1.

[0052] According to equation (41), the function can be expressed as (42) where if , then if , then .

[0053] Combining equation (36) and equation (41), the time derivative of the equivalent output variable can be described as (43) where (44) Rearranging equation (35) and equation (43), an algebraic expression about the lumped unknown input is obtained: (45) Step 3.2, according to equation (45), and using the first equation in Step 2, the equivalent output variable The observer provides an estimate of the warehouse temperature, and a lumped unknown input algebraic reconstruction mechanism is established to asymptotically reconstruct the lumped unknown input, so as to obtain an estimate of the unknown input in the model of the intelligent warehouse temperature control system.

[0054] First, it is noted that represents the time derivative of the function , which is unknown. Therefore, an estimate of is provided by using a sliding mode differentiator as follows: (46) where and represent the design parameters of the sliding mode differentiator, and are two states of the sliding mode differentiator, and the second state of the sliding mode differentiator will asymptotically converge to , i.e. , in a finite time.

[0055] For the lumped unknown input in the output equivalent transformed model (7) of the intelligent warehouse temperature control system, according to the algebraic relationship provided by equation (45), the derivative estimate provided by equation (46), and the estimate of the warehouse temperature provided by the distributed observer in step 2, a lumped unknown input algebraic reconstruction mechanism is established as follows: (47) where represents the reconstructed value of the lumped unknown input .

[0056] For the unknown input in the model (1) of the intelligent warehouse temperature control system, according to equation (4) and equation (47), the intelligent warehouse establishes an unknown input algebraic reconstruction mechanism as follows: (48) where represents the reconstructed value of the unknown input .

[0057] The verification process of the unknown input algebraic reconstruction mechanism for asymptotically reconstructing the unknown input is as follows: First, define the reconstruction error of the lumped unknown input as , and combine equation (45) and equation (47) to obtain (49) where , which is finite-time stable. It is noted that , and is bounded. Therefore, according to the distributed observer designed in Step 2 .

[0058] Then, define the reconstruction error of unknown input as According to equation (48) (50) Note that the matrix is a bounded constant matrix, therefore, according to the distributed observer designed in Step 2, the reconstruction error is asymptotically stable, satisfying , i.e., the unknown input reconstruction mechanism (48) can provide accurate unknown input reconstruction values.

[0059] Step 4, based on the temperature estimation value provided by the distributed observer in Step 2, and the lumped unknown input estimation value provided by the algebraic reconstruction mechanism of lumped unknown input in Step 3, combine the weighted average method to design a temperature adjustment controller to maintain the temperature of each partition of the intelligent warehouse at the desired value in the presence of unknown input.

[0060] For the intelligent warehouse temperature control system (1), in order to maintain the temperature of each partition at the desired value, according to the temperature estimation value and the unknown input estimation value , combine the weighted average method to design the following intelligent warehouse temperature adjustment controller: (51) where is the feedback gain matrix, which is selected to satisfy is a Hurwitz matrix, represents the temperature desired value, generated by the dynamic system .

[0061] The verification process of the temperature adjustment controller to ensure that the temperature of the intelligent warehouse is maintained at the desired value is as follows: Define the deviation between the temperature of the intelligent warehouse and the desired value as , combine equation (1) and equation (51), the time derivative of the deviation can be expressed as (52) Rearrange equation (17) and equation (52), the overall closed-loop system can be described as (53) where , , , , satisfy .

[0062] Note that the eigenvalues of matrix are the union of the eigenvalues of its block matrices and Moreover, matrices and are Hurwitz matrices. This shows that the overall closed-loop system is asymptotically stable, i.e., the designed temperature regulator guarantees that the temperature of the smart warehouse is maintained at the desired value.

[0063] Therefore, the temperature of the smart warehouse can be cooperatively estimated by the distributed observers, while the temperature of the smart warehouse can be maintained at the desired value under the action of the designed controller.

[0064] Simulation verification To verify the effectiveness of the proposed method, a smart warehouse with 9 partitions and a multi-sensor network consisting of 4 sensors are used for simulation verification. The partitioning and measurement information of the smart warehouse are shown in Figure 2 , where each box represents an independent partition; represents the control input generated by the heating device in the first partition, represents the control input generated by the cooling device in the fifth partition, represents the control input generated by the ventilation device in the ninth partition; represents the information that can be measured by the sensor from the first partition. In addition, the ninth partition is also affected by the external environment temperature.

[0065] The system matrix is chosen as (54) To complete the distributed monitoring and regulation of the temperature of the smart warehouse, the 4 sensors are respectively configured in the first, fifth, ninth, and third partitions, and the network communication topology structure among the 4 sensors is shown in Figure 3 , where the first, second, and third sensors can respectively access the input signals , , and in the first, fifth, and ninth partitions, while the fourth sensor cannot access any control input. Accordingly, the lumped unknown input and its coefficient matrix can be described as:

[0066] The measurement output information of each sensor is as follows:

[0067] The sensor output matrix is given as follows: - In addition, the unknown input in the smart warehouse temperature control system is chosen as , the sensor measurement noise is chosen as , and the coefficient matrix is chosen as , and The design objectives of the proposed method are as follows: first, a distributed observer is established based on the multi-sensor network to cooperatively estimate the temperature of the smart warehouse; then, an unknown input reconstruction mechanism based on the interval observer is constructed to achieve asymptotic estimation of the unknown input; finally, a controller for the smart warehouse temperature control system is established by combining the temperature estimation value provided by the distributed observer and the reconstructed value of the unknown input provided by the unknown input reconstruction mechanism, so that the temperature of each partition is stably maintained at the expected value. In the simulation, the expected temperature value of the smart warehouse is chosen as . Therefore, the controller is designed as (55) where the feedback gain is chosen such that the eigenvalues of the matrix are , the matrix is chosen as , and obviously .

[0068] To fully verify the feasibility and effectiveness of the proposed smart warehouse temperature distributed monitoring and regulation, a thermometer is arranged in each partition of the smart warehouse to record the real-time temperature change of the partition, and in addition, a thermometer is additionally arranged in the 9th partition to collect the real-time external environment temperature. The simulation results are shown in Figures 4-7 . Figure 4 shows the 2-norm trajectory of the deviation between the temperature estimation value provided by the distributed observer and the measured value by the thermometer. The results show that the 2-norm gradually converges to zero, verifying that the designed distributed observer can achieve cooperative monitoring of the temperature of the smart warehouse. Figure 5 depicts the trajectory of the adaptive law. In combination with Figure 4 , it can be observed that as the temperature estimation error converges, the adaptive law eventually stabilizes at a constant value. Figure 6 depicts the trajectory of the measured value of the thermometer arranged in the 9th partition (i.e., the unknown input signal) and its reconstructed value, from which it can be seen that the unknown input reconstruction can achieve asymptotic estimation of the unknown input signal. Figure 7 shows the measured value change trajectory of the temperature of each partition of the smart warehouse, and it can be observed that under the action of the temperature regulation controller, the temperature of all partitions of the smart warehouse will gradually be maintained at its expected value.

Claims

1. A temperature control method for an intelligent warehouse based on distributed monitoring, characterized by: The following steps are involved: S1: Based on the heat exchange mechanism between warehouse partitions and the outside world, combined with temperature measurement data collected by sensors evenly distributed in the warehouse, a smart warehouse temperature control system model with unknown inputs is constructed; S2: Based on the smart warehouse temperature control system model, we use detectability decoupling technology to build a distributed observer based on consistency theory to collaboratively estimate the temperature of each partition and obtain the temperature estimate of the smart warehouse; S3: To achieve compensatory control of unknown inputs, an interval observer is constructed for the output variables of the smart warehouse temperature control system model to obtain the algebraic relationship between the unknown input and output variables. Based on the estimated value of the smart warehouse temperature, an algebraic reconstruction mechanism for the unknown input is established to obtain the estimated value of the unknown input. S4: Based on the estimated value of the smart warehouse temperature and the estimated value of the unknown input, and combined with the weighted mean method and feedback control strategy, a smart warehouse temperature control controller is designed to maintain the temperature of the smart warehouse at the expected value in the presence of unknown input.

2. The intelligent warehouse temperature control method based on distributed monitoring according to claim 1 is characterized by: In step S1, the intelligent warehouse temperature control system model is: (1)； in, Indicates the temperature of the smart warehouse, Indicates the first The temperature of each zone, , is the number of partitions of the smart warehouse, ; represents the control input, Indicates the control input A quantity, , is a positive integer; Represents unknown input, used to describe external environmental disturbances or unmodeled heat source effects. is a positive integer; Represents the heat transfer coefficient matrix , Indicates the Partition and The heat exchange coefficient between the partitions, and represents a constant matrix; represents the temperature measurement output vector collected by sensors evenly spaced throughout the warehouse; Indicates the The measurement output vector collected by the sensors, , is the number of sensors; , is a positive integer; , Indicates the The measurement noise in each sensor channel, , is a positive integer; accordingly, the constant matrix and It can be expressed in blocks as and , therefore, for the sensors, whose measured output vector It can be expressed as: (2)； in and represents a constant matrix; Strongly connected graph Indicates that represents the set of sensor nodes, represents the sensor edge set, represents the sensor adjacency weight matrix, if A sensor can obtain the The information of each sensor, then the connection weight , if the The sensor cannot obtain the The information of each sensor, then the connection weight On Strongly Connected Graphs The Laplace matrix of Defined as ,in , .

3. The intelligent warehouse temperature control method based on distributed monitoring according to claim 2 is characterized by: Since each sensor node can only access part of the control input of the intelligent warehouse temperature control system model, sensors, It can be expressed as: (3)； in and Represents control input the accessible and inaccessible parts of and is a positive integer, satisfying , and is a constant matrix; Due to unknown input It is unknown for all sensors, so for the sensors, lumped unknown input and its coefficient matrix It can be described as: (4)； For the Sensors, combined with equations (1) to (4), the improved intelligent warehouse temperature control system model is: (5)； The coefficient matrix in formula (5) satisfies the following rank condition: (6)； in express dimensional identity matrix, .

4. The intelligent warehouse temperature control method based on distributed monitoring according to claim 3 is characterized by: Step S2 includes the following sub-steps: S2.1: In order to improve the intelligent warehouse temperature control system model, The negative impact of this, an equivalent output transformation is introduced ,in express dimensional identity matrix, Representation matrix The generalized inverse matrix of ; then, the improved intelligent warehouse temperature control system model is redescribed as: (7)； in ; Define the matrix for ,matrix and satisfy ,in Represents a matrix pair The undetectable subspace of Lemma 1: ; Lemma 1 shows that the following matrix equation has a solution, (8)； Its solution matrix The general form is ,in is an arbitrary constant matrix; Lemma 2: Definition , then the matrix pair and The undetectable subspace of is the same, that is, ； S2.2: Decouple the detectability of equation (7); let represents the undetectable subspace The dimensions of ; define the matrix , whose column vectors are represented by the undetectable subspace A set of orthogonal bases; at the same time, define the matrix , whose column vectors are represented by the undetectable subspace It is composed of a set of standard orthogonal bases of the orthogonal complement space of ; from this we can see that ,in Representation matrix The image space, Represents the matrix nuclear space; To obtain the matrix pair Decoupling the detectability of , constructing an orthogonal transformation matrix ; Due to the undetectable subspace It's about Invariant, then in the orthogonal transformation matrix Under the action of and Detectability breakdown: (9)； in , , , ; Matrix pair Is detectable, matrix It is unstable; S2.3: According to the equivalent output transformation in step S2.1 and the detectability decomposition in step S2.2, for equation (7), the distributed observer is: (10)； in, Indicates the The state of the observer, express estimated value of; Constant Matrix for: ; Feedback gain matrix for: , where the matrix Make is a Hurwitz matrix; Gain Matrix for: ; Coupling gain Designed for ,in is the adaptive law, The update rules are as follows: (11)； (12)； represents the initial moment, and is a positive design parameter.

5. The intelligent warehouse temperature control method based on distributed monitoring according to claim 4 is characterized by: Step S3 includes: S3.1: Due to the unknown input in Equation (1) Belongs to the actuator channel, that is , so there exists a matrix Make ; Accordingly, the matrix It can be expressed as ; In addition, in order to realize the lumped unknown input Reconstruction, there is at least one sensor node in the multi-sensor network, so that the matrix is of full rank; without loss of generality, assume that there exists The sensor nodes satisfy the matrix is a full-rank column; according to formula (7), we get the equivalent output variable The time derivative of : (35)； For equivalent output variables , the interval observer is: (36)； in and Represents the upper and lower bound states of the interval observer, whose initial conditions satisfy and , and Represents the initial conditions of the system The upper and lower bounds of , and Represents lumped unknown input The upper and lower bounds of ,matrix is a constant matrix that satisfies both the Hurwitz matrix condition and the Metzler matrix condition; there is a time constant , so that when When , the state of the interval observer and satisfy ; From formula (36), we can see that there is a time-varying function satisfy: (41)； in and , for any , The value range of is between 0 and 1; According to formula (41), the function It can be expressed as: (42)； in ,if ,So ,if ,So ; Combining equations (36) and (41), the equivalent output variables are The time derivative of can be described as: (43)； in, (44)； Arranging equations (35) and (43), we can obtain the lumped unknown input The algebraic expression for : (45)； Step 3.2: Since in formula (43) is unknown, the following sliding mode differentiator is used to provide Estimated value of : (46)； in and represents the design parameters of the sliding mode differentiator, and are the first and second states of the sliding mode differentiator, will converge asymptotically to ,Right now ; For the lumped unknown input in Eq. (7) , according to formula (45) and formula (46) The estimated value of and the estimated value of the smart warehouse temperature provided by the distributed observer in step 2 are used to establish the following lumped unknown input algebraic reconstruction mechanism: (47)； in Represents lumped unknown input The reconstruction value of For unknown input , according to equations (4) and (47), the unknown input algebraic reconstruction mechanism is established: (48)； in Indicates unknown input The reconstructed value of .

6. The intelligent warehouse temperature control method based on distributed monitoring according to claim 5 is characterized by: In step S4, the intelligent warehouse temperature adjustment controller is: (51)； in is the feedback gain matrix, which is selected to satisfy is the Hurwitz matrix, represents the expected value of temperature, which is given by the reference dynamic system whose state evolution satisfies ,in Indicates external input signal.

Images