9 results about "Evaluation function" patented technology

The invention provides an underwater tunnel shield construction excavation face stability evaluation method, system and equipment, and the method comprises the steps: determining an excavation face stability evaluation index based on a shield construction excavation face instability mechanism; dividing the stability of the excavation surface into a plurality of grades, establishing each grade space, and determining the quantitative interval of each evaluation index in each grade; calculating a combined weight of each stability evaluation index by adopting a combined weighting method, and taking the combined weight as a judgment basis of the influence of the evaluation indexes on an evaluation result; and constructing an ideal point evaluation function by adopting an ideal point method so as to represent the membership degree of the to-be-evaluated object to each grade, calculating the grade membership degree of the to-be-evaluated section in each grade of the evaluation system, and determining the stability grade of the excavation surface in the construction process of the evaluation section.

The invention provides a preparation method of a substrate of an external vesicle functional marker rapid detection kit, and belongs to the technical field of external vesicle functional marker detection.The preparation method comprises the steps that a high-purity coating working solution is prepared, and precise spraying is achieved through a piezoelectric type ink jet system; substrate functional modification is completed through confining liquid treatment and labeled antibody working solution preparation, a double-layer game model containing detection sensitivity optimization and process stability optimization is established, and an optimal parameter combination is obtained through game iterative calculation by combining the synergistic effect of a quality evaluation function and a stability control function. A characteristic parameter database is established to store historical production data and quality detection results, a multi-parameter joint evaluation method is adopted to calculate a comprehensive score and carry out unified product grading, and a quality control system and a standardized parameter optimization mechanism are driven through full-process data. The technical problem that the quality consistency difference is large in the batch production process is solved.

This invention proposes a priority-based hierarchical learning method. For N tasks with different priorities, the evaluation function for task i is denoted as Q. i and maintain a prediction model π i For task i with priority, and all higher priority tasks j < i, prepare a predetermined threshold ε. ij and maintain a slack variable λ ij For any i > j, if Q j (π i )+ε ij <Q j (π j ), then represents π. i Performance on task j was better than π. j If the difference is too large, then increase λ. ij Conversely, λ decreases. ij But keep λ ij >0. With λ ij j < i is used as the weight to optimize π i Repeat the previous steps until convergence, and finally obtain π. N This is the desired model. In this invention, constraints are used to describe priorities, solving the problem of traditional multi-objective optimization methods lacking a priority order. Slack variables are introduced to automatically adjust the weights of each optimization objective. The dual variable is adaptively adjusted, resulting in zero duality with the primal problem, making it a convex optimization problem that can be solved quickly by existing solvers.

The invention provides a task processing method and device, electronic equipment and a computer storage medium, and relates to the technical fields of automatic driving, artificial intelligence and the like. According to the specific implementation scheme, based on a multi-dimensional task value constraint file loaded by a user, a value evaluation rule set is extracted; acquiring state sequence data of the monitored object in the monitoring time period; obtaining a value evaluation function set and a function boundary condition set of the value evaluation function set based on the task description information of the target task, the value evaluation rule set and the state sequence data; receiving task target information of the target task; based on the task target information, the value evaluation function set and the function boundary condition set, a target decision model is constructed, and the target decision model takes a task utility function corresponding to the task target information as an optimization target and takes the value evaluation function set and the function boundary condition set as constraint conditions; and obtaining a target optimization result of the target task based on the target decision model.

The invention provides a control method for heat release of a heat storage power station, and belongs to the technical field of heat storage power stations, and the method comprises the steps: collecting the comprehensive state parameters of a heat storage medium, building a database, and constructing an upper layer game model with power grid load matching as a target and a lower layer game model with heat storage medium stability as a target; a thermal load prediction optimization model based on a Performmer architecture is used for prediction control, a variable frequency driving circulating pump control system is established to achieve accurate flow adjustment, and the optimal flow distribution proportion of different temperature layers is calculated through a heat storage medium layered optimization function. A coordination control strategy of a high-temperature layer priority heat release mode, a medium-temperature layer stable heat release mode and a low-temperature layer supplementary heat release mode is adopted, control parameters are dynamically adjusted through a heat release process stability evaluation function, and the technical problem that stability control over heat release of the heat storage power station and power grid load matching are poor in coordination is solved.

Disclosed herein is an optimization device, an optimization method, and an optimization program. The optimization device includes: an annealing unit that changes a state of any one of a plurality of state variables included in an evaluation function, calculates an amount of energy change, and obtains a first total amount of change by adding a second total amount of change to the calculated amount of change; a speculative inversion control unit that repeatedly performs a process of speculatively selecting a state variable to be changed and causing the annealing unit to obtain the first total amount of change; an adoption determination unit that randomly determines whether to adopt a state transition of a predetermined number of state variables changed by the annealing unit; an energy calculation unit that calculates a transition energy when it is determined to adopt the state transition; and a search unit that specifies the transition energy as a minimum energy when the transition energy is smaller than a previously specified minimum energy.

The invention provides a grouping relative strategy optimization method with stable dynamic reference, which comprises the following steps of: performing multiple independent sampling on the same input prompt by a current strategy model, generating a candidate answer set, calculating an absolute quality score of each candidate answer through a preset determinacy evaluation function, and calculating a quality score of each candidate answer; calculating an average score of all candidate answers in the current group; dynamically updating a global historical reference value according to the score of the current batch of candidate answers through a dynamic global historical reference line maintenance module; through a composite advantage calculation module, in-group relative comparison and global historical comparison are fused, and a composite advantage value of each candidate answer is calculated; and through a strategy gradient updating module, the logarithmic probability gradient of the strategy model is calculated and the parameters of the strategy model are updated by taking the composite advantage value as a weighted weight, so that the generation probability of answers with high advantage values is enhanced, and the generation probability of answers with low advantage values is inhibited. The method has the beneficial effect that the answer which is absolutely progressive compared with the historical performance of the model can be accurately identified.

The invention discloses a double-party game training method and system based on a reinforcement learning algorithm, and a storage medium, and belongs to the technical field of game control. According to the two-party game training method based on the reinforcement learning algorithm, the state information of the two game parties is obtained to serve as input parameters of a fighting model, a situation evaluation function and a reward function are combined, and a height limiting condition and a speed limiting condition are set, so that a control model based on the reinforcement learning algorithm is constructed; the model is fully trained by presetting the upper limit value of the training parameter, so that a higher situation assessment score is obtained, and the prediction accuracy of situation development is improved.

Efficiently search for solutions to mixed integer programming problems. [Solution] The processing unit 12 determines the first solution of an alternative problem, which is represented by a second evaluation function obtained by adding the product of an auxiliary variable corresponding to the constraints and a weight coefficient to the first evaluation function of a mixed integer programming problem containing integer variables and continuous variables, using the branch and bound method with the integer variables linearly relaxed. The processing unit 12 then determines the second solution of the mixed integer programming problem by local search, using an initial solution in which the linearly relaxed integer variables in the first solution are set to integer values, while fixing the values ​​of the continuous variables included in the initial solution. The processing unit 12 then determines the third solution of the mixed integer programming problem with the values ​​of the integer variables included in the second solution fixed, and decreases the values ​​of the weight coefficients corresponding to the constraints satisfied by the third solution. The processing unit 12 then repeats the process of determining the first solution, determining the second solution, determining the third solution, and decreasing the values ​​of the weight coefficients.